{"meta":{"query_hash":"12114726b7e6","filters":{"venue":"Biostatistics"},"cohort_total":82,"direct_labels_cover":0,"predictions_cover":82,"exported":82,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/12114726b7e6","api":"https://metacan.xera.ac/api/v1/cohort?venue=Biostatistics"},"results":[{"id":"W1528102277","doi":"10.1093/biostatistics/kxl026","title":"Cluster-based network model for time-course gene expression data","year":2006,"lang":"en","type":"article","venue":"Biostatistics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital; University of British Columbia","funders":"National Cancer Institute","keywords":"Cluster analysis; Computer science; Data mining; Gene cluster; Computational biology; Gene regulatory network; Expression (computer science); Cluster (spacecraft); Gene expression; Gene; Artificial intelligence; Biology; Genetics; Computer network","score_opus":0.026755659162620793,"score_gpt":0.29276575341318256,"score_spread":0.26601009425056177,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1528102277","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009903731,0.000109336645,0.98770857,0.00029271,0.000024401888,0.00006670637,0.00064752146,0.00031533174,0.00093176606],"genre_scores_gemma":[0.54953,0.0010996124,0.42788938,0.00042514974,0.0001695573,0.0019399162,0.0053207134,0.00034896634,0.013276589],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99846846,0.00054749724,0.00006635735,0.0004725083,0.00032864392,0.000116622825],"domain_scores_gemma":[0.9961998,0.0024689918,0.0004315647,0.00030713915,0.00046506996,0.00012750241],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028253132,0.0010380608,0.0013046279,0.002150324,0.00066558557,0.0014513002,0.003358953,0.001546692,0.0032353573],"category_scores_gemma":[0.00797755,0.00054347044,0.0014507606,0.0024827833,0.0011428616,0.0020860427,0.001152083,0.0018570378,0.0010250073],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006400385,0.0000341778,0.0011444921,0.00004459437,0.00008017622,0.000060576254,0.00008630157,0.9214747,0.0009846981,0.06520346,0.0010514827,0.0097713815],"study_design_scores_gemma":[0.0000051768693,0.000006751833,0.00015788064,0.0000025462532,0.00000773149,0.000012758248,0.0000046762207,0.9767553,0.00009080504,0.022476047,0.00047451156,0.0000058817677],"about_ca_topic_score_codex":0.012850167,"about_ca_topic_score_gemma":0.012443136,"teacher_disagreement_score":0.012850167,"about_ca_system_score_codex":0.0023246838,"about_ca_system_score_gemma":0.0014025237,"threshold_uncertainty_score":0.025550723},"labels":[],"label_agreement":null},{"id":"W1994381110","doi":"10.1093/biostatistics/kxl035","title":"A hierarchical clustering method for estimating copy number variation","year":2006,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genomic variations and chromosomal abnormalities","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Hospital for Sick Children; Lunenfeld-Tanenbaum Research Institute; SickKids Foundation; Mount Sinai Hospital","funders":"National Heart, Lung, and Blood Institute; Canadian Institutes of Health Research","keywords":"Cluster analysis; Outlier; Computer science; Statistic; Tree (set theory); Change detection; Data mining; Scan statistic; Hierarchical clustering; Copy-number variation; Statistics; Pattern recognition (psychology); Mathematics; Artificial intelligence; Genome; Biology; Genetics","score_opus":0.009536162042363535,"score_gpt":0.2802942228825624,"score_spread":0.2707580608401989,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1994381110","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016618706,0.00011005097,0.99656314,0.000045196153,0.000029740624,0.00008837668,0.00023513437,0.00091241073,0.00035411035],"genre_scores_gemma":[0.031397983,0.00013360305,0.96564776,0.000058372643,0.00005148691,0.00030847074,0.00083455985,0.00024337273,0.0013243156],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9967404,0.0009856104,0.00015698042,0.00065257546,0.0013133516,0.00015113196],"domain_scores_gemma":[0.99701023,0.001285866,0.00028905235,0.0005091892,0.000806175,0.00009947287],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028436074,0.0012379739,0.0012787337,0.004739564,0.0013103811,0.0010578119,0.002398161,0.0010528865,0.003221253],"category_scores_gemma":[0.008817458,0.00065957877,0.0018476967,0.0046479483,0.00061325793,0.0009014656,0.0010430585,0.0014914705,0.0021309704],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015278501,0.00012202211,0.005353616,0.00033240646,0.0007501851,0.00020379429,0.00040846586,0.13681918,0.016932232,0.030193109,0.014507803,0.79422444],"study_design_scores_gemma":[0.000051145606,0.000095781754,0.006554487,0.00003948809,0.00013465412,0.0004390588,0.00007671234,0.93068373,0.008369095,0.034629315,0.018735087,0.00019137078],"about_ca_topic_score_codex":0.014484386,"about_ca_topic_score_gemma":0.013984034,"teacher_disagreement_score":0.014484386,"about_ca_system_score_codex":0.0013498474,"about_ca_system_score_gemma":0.0025339576,"threshold_uncertainty_score":0.02880013},"labels":[],"label_agreement":null},{"id":"W2011218435","doi":"10.1093/biostatistics/kxt052","title":"Analysis of counts with two latent classes, with application to risk assessment based on physician-visit records of cancer survivors","year":2013,"lang":"en","type":"article","venue":"Biostatistics","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency; Simon Fraser University","funders":"BC Cancer Agency","keywords":"Estimator; Statistics; Consistency (knowledge bases); Unobservable; Normality; Variance (accounting); Inference; Mathematics; Latent class model; Asymptotic distribution; Econometrics; Delta method; Computer science; Artificial intelligence","score_opus":0.009125071757745046,"score_gpt":0.2977986433723875,"score_spread":0.28867357161464247,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2011218435","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.56132406,0.0004235029,0.43562245,0.0009705164,0.00001873191,0.00018010668,0.00034605208,0.00018312677,0.0009314519],"genre_scores_gemma":[0.8897329,0.00019750485,0.10827768,0.000045199635,0.000044046265,0.0002208522,0.0005022227,0.00003377779,0.000945837],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99189055,0.005871613,0.0002582246,0.0006802284,0.0010921678,0.00020716703],"domain_scores_gemma":[0.88934225,0.09801878,0.006631189,0.003499392,0.0018769994,0.0006313327],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.016014105,0.00055481296,0.0011474731,0.0022606452,0.0008489859,0.0019111698,0.0018665618,0.0014292614,0.001319416],"category_scores_gemma":[0.09606917,0.0005637471,0.0010773106,0.002537308,0.0017733596,0.0021707553,0.0022626605,0.0019930643,0.00016512054],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001036366,0.0008172077,0.36980957,0.00032616328,0.00052638655,0.0005797104,0.0036028505,0.26717055,0.0019381632,0.1511931,0.001802251,0.20119767],"study_design_scores_gemma":[0.00004156721,0.00011302719,0.03380512,0.000032118664,0.000042644115,0.00015071922,0.0005414019,0.91968673,0.0005921168,0.04424458,0.00070214353,0.00004776657],"about_ca_topic_score_codex":0.0076029147,"about_ca_topic_score_gemma":0.0065297154,"teacher_disagreement_score":0.016014105,"about_ca_system_score_codex":0.0013867036,"about_ca_system_score_gemma":0.001208814,"threshold_uncertainty_score":0.08469164},"labels":[],"label_agreement":null},{"id":"W2015295476","doi":"10.1093/biostatistics/kxt054","title":"Efficient panel designs for longitudinal recurrent event studies recording panel counts","year":2013,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Centre for Disease Control; Western University","funders":"Western Canada Research Grid; Compute Canada","keywords":"Panel data; Covariate; Computer science; Outcome (game theory); Event (particle physics); Econometrics; Panel analysis; Statistics; Set (abstract data type); Piecewise; Mathematics","score_opus":0.8361934892466871,"score_gpt":0.5864674015450662,"score_spread":0.24972608770162097,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2015295476","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019044242,0.00023489076,0.9966402,0.00018078687,0.00004396843,0.000359658,0.00018788625,0.00009318838,0.0003550955],"genre_scores_gemma":[0.067699455,0.0009378829,0.92085046,0.0004401955,0.00022981477,0.007233377,0.0009778171,0.0001374761,0.0014933618],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.84275866,0.14316814,0.0033251818,0.0055107083,0.004130027,0.0011073537],"domain_scores_gemma":[0.6515493,0.28837976,0.016812976,0.03485659,0.0070409905,0.0013604325],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.15656158,0.0022892514,0.0039562923,0.0020860576,0.00097507256,0.0029905732,0.0051790294,0.0040024673,0.008430347],"category_scores_gemma":[0.27474362,0.0022540232,0.0037089984,0.0030830763,0.0035265756,0.0048054475,0.0044061122,0.0063288915,0.0015758069],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013891744,0.00030186056,0.0051285303,0.0012782196,0.0011084562,0.00028342195,0.0012163124,0.13623174,0.0022890163,0.7361893,0.004567767,0.11001617],"study_design_scores_gemma":[0.0008549847,0.0011613328,0.0022102864,0.00040530055,0.0004679924,0.00015666007,0.00015341857,0.29629803,0.002131273,0.6843177,0.011718102,0.000125024],"about_ca_topic_score_codex":0.0009746216,"about_ca_topic_score_gemma":0.0010949449,"teacher_disagreement_score":0.15656158,"about_ca_system_score_codex":0.0017794188,"about_ca_system_score_gemma":0.0034161296,"threshold_uncertainty_score":0.8279867},"labels":[],"label_agreement":null},{"id":"W2095702029","doi":"10.1093/biostatistics/kxn034","title":"Exact and efficient inference procedure for meta-analysis and its application to the analysis of independent 2 x 2 tables with all available data but without artificial continuity correction","year":2008,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":123,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Group for Research in Decision Analysis; Université du Québec à Montréal","funders":"U.S. National Library of Medicine; National Institute of Allergy and Infectious Diseases; National Heart, Lung, and Blood Institute","keywords":"Computer science; Inference; Simple (philosophy); Meta-analysis; Data mining; Continuity correction; Statistics; Econometrics; Mathematics; Artificial intelligence; Medicine; Internal medicine","score_opus":0.5257705219849328,"score_gpt":0.5038448131116642,"score_spread":0.021925708873268612,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2095702029","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00044035361,0.00018919038,0.9985316,0.0001301722,0.000057205427,0.00012180802,0.00013451146,0.000264733,0.00013046674],"genre_scores_gemma":[0.015023098,0.00028824204,0.9823184,0.00014478552,0.00012916022,0.0013374339,0.00020836531,0.00014380064,0.00040676293],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.93985826,0.051811874,0.0022852386,0.002259133,0.003454572,0.00033097912],"domain_scores_gemma":[0.8256739,0.15675922,0.002834229,0.011337029,0.0030457191,0.00034992548],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.06095853,0.0020535355,0.0036345462,0.0048842905,0.0010829701,0.001477637,0.0032070165,0.0016995855,0.010315458],"category_scores_gemma":[0.19741611,0.0015801867,0.0041402783,0.00651017,0.001694951,0.0022880482,0.00261765,0.0057125585,0.0016378227],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001082539,0.00026542525,0.004318588,0.0021664442,0.0063960194,0.0010140686,0.00077751296,0.083384134,0.0027448745,0.20366657,0.020166408,0.67401737],"study_design_scores_gemma":[0.0011030098,0.0006547585,0.0035790661,0.00036591012,0.0020406765,0.0010657761,0.00011414448,0.47402954,0.0040595233,0.48687488,0.02588462,0.00022803036],"about_ca_topic_score_codex":0.0027625142,"about_ca_topic_score_gemma":0.0028552401,"teacher_disagreement_score":0.9390415,"about_ca_system_score_codex":0.0010043087,"about_ca_system_score_gemma":0.0045753275,"threshold_uncertainty_score":0.3223834},"labels":[],"label_agreement":null},{"id":"W2095887717","doi":"10.1093/biostatistics/3.2.229","title":"A Bayesian approach to case-control studies with errors in covariables","year":2002,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bayesian probability; Computer science; Control (management); Econometrics; Statistics; Artificial intelligence; Mathematics","score_opus":0.1371729806175621,"score_gpt":0.361720640302282,"score_spread":0.2245476596847199,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2095887717","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003543695,0.0007951961,0.9975979,0.0006528398,0.00005823899,0.00010737195,0.00004934719,0.000032756932,0.00035196892],"genre_scores_gemma":[0.03016776,0.0033156816,0.9613918,0.00090580445,0.00055386615,0.0024547433,0.00020395272,0.00005189614,0.0009544241],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.92945206,0.056132935,0.003122569,0.0034881015,0.007259633,0.0005446374],"domain_scores_gemma":[0.882722,0.10346285,0.004541818,0.0056029116,0.0029287555,0.0007417179],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.07404706,0.0019420581,0.0039443295,0.0072669187,0.0015068464,0.0036395935,0.006572803,0.0036429898,0.003962041],"category_scores_gemma":[0.1872464,0.0017569319,0.0029112466,0.0063818316,0.0040647774,0.0033503822,0.003909808,0.0069149206,0.00080807396],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006496746,0.000107704174,0.0020999475,0.00056473276,0.00073742686,0.00043789018,0.0006012455,0.04144092,0.00049523753,0.8673378,0.003496136,0.08261585],"study_design_scores_gemma":[0.00009434753,0.0000863095,0.00047308183,0.00016912879,0.00016491655,0.0002988284,0.000037288402,0.05430382,0.00016532914,0.9364691,0.007692927,0.000044908276],"about_ca_topic_score_codex":0.0048613977,"about_ca_topic_score_gemma":0.004660575,"teacher_disagreement_score":0.07404706,"about_ca_system_score_codex":0.0026643733,"about_ca_system_score_gemma":0.0050003226,"threshold_uncertainty_score":0.39160293},"labels":[],"label_agreement":null},{"id":"W2097681968","doi":"10.1093/biostatistics/2.3.323","title":"Efficiency considerations in the analysis of inter-observer agreement","year":2001,"lang":"en","type":"article","venue":"Biostatistics","topic":"Reliability and Agreement in Measurement","field":"Decision Sciences","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Cohen's kappa; Replicate; Statistics; Inter-rater reliability; Reliability (semiconductor); Statistic; Kappa; Mathematics; Binary number; Computer science; Binomial distribution; Econometrics; Arithmetic; Physics","score_opus":0.23113506979270315,"score_gpt":0.40521817031104557,"score_spread":0.17408310051834242,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097681968","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0049557583,0.00068680185,0.9924064,0.0003754382,0.00007683824,0.00030403532,0.0000650285,0.00011769133,0.0010118803],"genre_scores_gemma":[0.31867346,0.0008259462,0.67255867,0.0010571405,0.0003105336,0.0029624768,0.0004508146,0.0007040234,0.0024568776],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.47599486,0.4531517,0.01780426,0.020285651,0.030619312,0.0021441593],"domain_scores_gemma":[0.13474303,0.80217546,0.014272391,0.035514615,0.012876503,0.00041796148],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3907782,0.002745539,0.0045596203,0.0041220183,0.0017419736,0.005607869,0.005316505,0.004008732,0.002050049],"category_scores_gemma":[0.68337536,0.0019559928,0.0048409947,0.0050296546,0.0079199895,0.008073524,0.007114369,0.0057766573,0.0013342503],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018385944,0.00041432044,0.04883049,0.002479892,0.0064648762,0.0011131536,0.0064405384,0.15859601,0.0049408004,0.46095097,0.006433282,0.30149707],"study_design_scores_gemma":[0.00022142704,0.0010678046,0.027752012,0.0009114322,0.0008723466,0.0013040707,0.0009919952,0.45852253,0.007953979,0.48084658,0.01923726,0.00031858563],"about_ca_topic_score_codex":0.0025977904,"about_ca_topic_score_gemma":0.0015171148,"teacher_disagreement_score":0.6092218,"about_ca_system_score_codex":0.0040751314,"about_ca_system_score_gemma":0.0033665453,"threshold_uncertainty_score":0.75127894},"labels":[],"label_agreement":null},{"id":"W2098860533","doi":"10.1093/biostatistics/kxq016","title":"Surface shape analysis with an application to brain surface asymmetry in schizophrenia","year":2010,"lang":"en","type":"article","venue":"Biostatistics","topic":"Morphological variations and asymmetry","field":"Mathematics","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Asymmetry; Surface (topology); Bayesian probability; Shape analysis (program analysis); Mathematics; Artificial intelligence; Pattern recognition (psychology); Computer science; Principal component analysis; Scale (ratio); Geometry; Physics","score_opus":0.02034049229156815,"score_gpt":0.30548675897984356,"score_spread":0.2851462666882754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098860533","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.038008884,0.00083868,0.9584913,0.00037805983,0.00005664149,0.00011192728,0.00033332067,0.0006377202,0.0011434217],"genre_scores_gemma":[0.26190916,0.0011380686,0.7347844,0.00005833419,0.00016170257,0.00023638892,0.0004206006,0.00027459857,0.0010167311],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9993994,0.00022126065,0.00006293245,0.00006357544,0.00021278305,0.00004008899],"domain_scores_gemma":[0.99870646,0.00068379065,0.00018052458,0.00018391333,0.0001963598,0.000049063787],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001635993,0.0005848938,0.00063824607,0.004013147,0.0003855693,0.000740156,0.00061718596,0.00054669264,0.0023746034],"category_scores_gemma":[0.0047168424,0.00030845773,0.0010889336,0.00282142,0.00097222615,0.00045543897,0.00079789694,0.00092627667,0.0004273568],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031523922,0.00007694689,0.016775297,0.00039827745,0.00020493142,0.0012777238,0.00085853686,0.07862902,0.07266372,0.050719842,0.0032017438,0.7748787],"study_design_scores_gemma":[0.00005500868,0.00031864332,0.075548895,0.00008620868,0.00011169955,0.0052004023,0.0006374426,0.73076785,0.022208817,0.14869003,0.016208844,0.00016629205],"about_ca_topic_score_codex":0.0017881214,"about_ca_topic_score_gemma":0.0011342684,"teacher_disagreement_score":0.004013147,"about_ca_system_score_codex":0.00038362882,"about_ca_system_score_gemma":0.0003793967,"threshold_uncertainty_score":0.008652091},"labels":[],"label_agreement":null},{"id":"W2100420476","doi":"10.1093/biostatistics/kxr010","title":"Classifying tissue samples from measurements on cells with within-class tissue sample heterogeneity","year":2011,"lang":"en","type":"article","venue":"Biostatistics","topic":"AI in cancer detection","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"BC Cancer Agency","funders":"National Cancer Institute","keywords":"Latent class model; Computer science; Macro; Replicate; Class (philosophy); Heuristic; Object (grammar); Focus (optics); Sample (material); Cluster analysis; Statistics; Mixture model; Artificial intelligence; Pattern recognition (psychology); Mathematics; Machine learning","score_opus":0.13299273575744242,"score_gpt":0.2858008119355312,"score_spread":0.15280807617808878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100420476","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2507005,0.0006783667,0.74651635,0.0004892655,0.000074406125,0.00013982064,0.0003312619,0.00034344886,0.00072665006],"genre_scores_gemma":[0.81124395,0.00035033285,0.18631607,0.00025712472,0.00012141705,0.00017862732,0.00073871174,0.00004288556,0.00075080787],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99802566,0.00046098963,0.00020539037,0.0005672311,0.00060969713,0.00013111393],"domain_scores_gemma":[0.9932226,0.004391662,0.0009096277,0.000969389,0.00040424097,0.00010245865],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0034514891,0.00049245334,0.0011273108,0.0015888262,0.00034724633,0.0014144442,0.0009704803,0.0015196357,0.0006361324],"category_scores_gemma":[0.0129293185,0.00033138192,0.0008886941,0.0013688437,0.0012237442,0.0011993336,0.0007835526,0.00077387696,0.0002956266],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00091668114,0.00034830757,0.2180752,0.00094203465,0.0005060557,0.0015421169,0.0009964596,0.19776319,0.1146407,0.02197263,0.0021131637,0.44018352],"study_design_scores_gemma":[0.000051940486,0.00027177314,0.08575671,0.00006706564,0.0001587826,0.0017059523,0.0005767071,0.80953264,0.03607642,0.06154322,0.004172994,0.00008576034],"about_ca_topic_score_codex":0.0016957686,"about_ca_topic_score_gemma":0.0016633541,"teacher_disagreement_score":0.0034514891,"about_ca_system_score_codex":0.00059913227,"about_ca_system_score_gemma":0.00052223005,"threshold_uncertainty_score":0.018253505},"labels":[],"label_agreement":null},{"id":"W2102283511","doi":"10.1093/biostatistics/kxv007","title":"Hypothesis testing at the extremes: fast and robust association for high-throughput data","year":2015,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Cancer Institute; National Heart, Lung, and Blood Institute; National Institute of Mental Health; Ontario Genomics Institute; Cystic Fibrosis Canada; National Institute of Environmental Health Sciences; Genome Canada; Ontario Genomics; Cystic Fibrosis Foundation","keywords":"Computer science; Throughput; Association (psychology); Data mining; Statistics; Mathematics; Psychology","score_opus":0.1521403879759942,"score_gpt":0.3005252873365745,"score_spread":0.14838489936058033,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102283511","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001991557,0.0005876917,0.99261945,0.000300864,0.00008780342,0.00011497796,0.00043147441,0.0034443801,0.00042190964],"genre_scores_gemma":[0.029770803,0.00066274987,0.96470803,0.00036783406,0.0001693651,0.001007137,0.0013751406,0.0009116083,0.0010272911],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.97370094,0.017001024,0.0012328061,0.0032869033,0.004151496,0.000626737],"domain_scores_gemma":[0.8587571,0.11705638,0.004944683,0.013429979,0.0041209506,0.0016909476],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04481686,0.002202696,0.004789368,0.0035314013,0.001389927,0.0042791613,0.0058520855,0.0036684894,0.008637732],"category_scores_gemma":[0.16393165,0.0023424805,0.0038307125,0.0056786165,0.0029074703,0.004514946,0.006340775,0.0070831995,0.0072503667],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002149248,0.0003850394,0.024031216,0.002538614,0.0027957032,0.002402458,0.0010964129,0.16116256,0.01792218,0.1031994,0.051106676,0.63121057],"study_design_scores_gemma":[0.0003456004,0.0003520704,0.00648602,0.0002502596,0.00033373607,0.0015235305,0.00016160654,0.6917759,0.0072504436,0.27033845,0.020925473,0.00025688988],"about_ca_topic_score_codex":0.0025207957,"about_ca_topic_score_gemma":0.0033680848,"teacher_disagreement_score":0.04481686,"about_ca_system_score_codex":0.0013116329,"about_ca_system_score_gemma":0.003076484,"threshold_uncertainty_score":0.2370171},"labels":[],"label_agreement":null},{"id":"W2105004777","doi":"10.1093/biostatistics/kxr005","title":"Weighted scores method for regression models with dependent data","year":2011,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Univariate; Copula (linguistics); Statistics; Mathematics; Weighting; Multivariate statistics; Negative binomial distribution; Regression; Regression analysis; Inference; Binomial regression; Econometrics; Computer science; Poisson distribution; Artificial intelligence; Medicine","score_opus":0.2664853473612795,"score_gpt":0.4255545416501072,"score_spread":0.1590691942888277,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2105004777","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007203348,0.00008242907,0.99866366,0.000051645213,0.000020094074,0.000027916778,0.00004730683,0.00009548016,0.00029111514],"genre_scores_gemma":[0.05664639,0.000710716,0.93555164,0.0002120485,0.00023910064,0.0009061335,0.0007743999,0.00044322826,0.004516315],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99044377,0.0063949646,0.00031523095,0.0008094515,0.0018248762,0.00021163846],"domain_scores_gemma":[0.978182,0.016342882,0.00125613,0.002118903,0.001723302,0.00037682016],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013171795,0.0018425155,0.001865671,0.004126997,0.00076610106,0.0019518296,0.0043984856,0.0017947819,0.00897602],"category_scores_gemma":[0.049360428,0.00084976776,0.0021019282,0.0039767437,0.002276632,0.0041642543,0.0028990214,0.003992062,0.0023152805],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008675212,0.000101367616,0.0018169084,0.0002702397,0.0002800278,0.00018612594,0.00020933549,0.11292373,0.0012983292,0.6932161,0.004654196,0.18495688],"study_design_scores_gemma":[0.00002879053,0.00006028924,0.0005424573,0.000049978822,0.000047143458,0.00011470057,0.0000387043,0.6116886,0.0007387042,0.37668297,0.0099661,0.000041555817],"about_ca_topic_score_codex":0.0037940913,"about_ca_topic_score_gemma":0.0033727905,"teacher_disagreement_score":0.013171795,"about_ca_system_score_codex":0.0011754245,"about_ca_system_score_gemma":0.0023938266,"threshold_uncertainty_score":0.06965995},"labels":[],"label_agreement":null},{"id":"W2107365898","doi":"10.1093/biostatistics/kxp029","title":"Second-order estimating equations for the analysis of clustered current status data","year":2009,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Covariate; Copula (linguistics); Marginal model; Quantile; Marginal distribution; Estimating equations; Statistics; Generalized estimating equation; Econometrics; Mathematics; Joint probability distribution; Order statistic; Regression analysis; Maximum likelihood","score_opus":0.3129426196990909,"score_gpt":0.48874706451441663,"score_spread":0.17580444481532576,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107365898","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014557388,0.0002340315,0.9973515,0.00019012079,0.000026794087,0.000077325094,0.00031088668,0.0001463331,0.00020720204],"genre_scores_gemma":[0.06584115,0.0018910742,0.9236199,0.00031110275,0.00016901545,0.0017662821,0.0021808313,0.00020426408,0.0040164026],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98508424,0.011240688,0.0007614303,0.0013151438,0.0013109288,0.00028757023],"domain_scores_gemma":[0.9156715,0.07183876,0.0039631883,0.005393075,0.0028802631,0.00025318615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02883622,0.0014018487,0.0022650526,0.002150762,0.0006114324,0.001653891,0.0025763272,0.0015237537,0.007324106],"category_scores_gemma":[0.103407875,0.0012608003,0.0026417647,0.0038420307,0.0010668046,0.0023654103,0.0017606777,0.004693556,0.0017959987],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008167089,0.00014589101,0.01714302,0.00066513,0.0011585143,0.00028213178,0.0009781226,0.11729902,0.0010226895,0.67865044,0.011250364,0.17132303],"study_design_scores_gemma":[0.00008227031,0.00011575956,0.0053461203,0.00015573602,0.0002232812,0.00020922664,0.000121721394,0.544523,0.00072720024,0.4335054,0.014908268,0.00008199186],"about_ca_topic_score_codex":0.012451998,"about_ca_topic_score_gemma":0.013668344,"teacher_disagreement_score":0.02883622,"about_ca_system_score_codex":0.00178155,"about_ca_system_score_gemma":0.0040404657,"threshold_uncertainty_score":0.15250236},"labels":[],"label_agreement":null},{"id":"W2107721271","doi":"10.1093/biostatistics/kxv012","title":"Semiparametric likelihood inference for left-truncated and right-censored data","year":2015,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Health Canada; Medical Research Council; Pfizer Canada; University of Ottawa; National Cancer Institute; National Institutes of Health; Pfizer","keywords":"Mathematics; Censoring (clinical trials); Estimator; Truncation (statistics); Multiplicative function; Statistics; Semiparametric model; Parametric statistics; Test statistic; Parametric model; Applied mathematics; Statistical hypothesis testing","score_opus":0.20177109706344662,"score_gpt":0.42213631169961596,"score_spread":0.22036521463616934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2107721271","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002520687,0.00006082711,0.99713147,0.00004725209,0.000007056105,0.000008896627,0.00002752027,0.00006299891,0.00013316495],"genre_scores_gemma":[0.29159757,0.00060452794,0.7045364,0.00017574838,0.00013721148,0.00028101527,0.0006619544,0.000149629,0.0018559598],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99127805,0.0059192926,0.00043790755,0.00093255023,0.0012153976,0.00021676539],"domain_scores_gemma":[0.9449856,0.046623677,0.0029699097,0.0039488184,0.0011915005,0.00028047743],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0140212765,0.0007695155,0.0017166319,0.0020293002,0.0004933317,0.0016081793,0.0028402335,0.0010965478,0.001665564],"category_scores_gemma":[0.084260486,0.00077110546,0.0018554096,0.0016906607,0.002109233,0.002635741,0.0028351655,0.0026308917,0.00045633918],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018611792,0.00017174512,0.0097180335,0.00037701195,0.0004452213,0.00041119475,0.0006071839,0.31901884,0.0023666564,0.459478,0.001575622,0.20564431],"study_design_scores_gemma":[0.00003203076,0.000048520596,0.0012813098,0.000046166624,0.00004346684,0.00014020379,0.000045048608,0.8136242,0.0011688191,0.18227835,0.0012520759,0.000039779297],"about_ca_topic_score_codex":0.002001558,"about_ca_topic_score_gemma":0.0016927731,"teacher_disagreement_score":0.0140212765,"about_ca_system_score_codex":0.0010896397,"about_ca_system_score_gemma":0.0018537597,"threshold_uncertainty_score":0.07415247},"labels":[],"label_agreement":null},{"id":"W2109802067","doi":"10.1093/biostatistics/1.4.441","title":"Monitoring surgical performance using risk-adjusted cumulative sum charts","year":2000,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Statistical Process Monitoring","field":"Decision Sciences","cited_by":473,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"CUSUM; Context (archaeology); Computer science; Quality (philosophy); Surgical procedures; Statistics; Variable (mathematics); Medicine; Surgery; Mathematics","score_opus":0.19504895291578445,"score_gpt":0.43317991759428615,"score_spread":0.2381309646785017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109802067","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48649698,0.0011584059,0.49132627,0.0004808474,0.00013620709,0.00080469897,0.0064063114,0.007117089,0.0060731354],"genre_scores_gemma":[0.8594382,0.00045111668,0.1357451,0.00005174779,0.00010786483,0.00038338042,0.0029104515,0.00013283298,0.0007793664],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9919117,0.0034642036,0.00091187266,0.0008025037,0.00269121,0.00021844092],"domain_scores_gemma":[0.9588704,0.020608809,0.00979324,0.0032592323,0.0063957977,0.0010725614],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0071081603,0.0008122296,0.0008751024,0.00550211,0.0003246509,0.0012742722,0.001121974,0.00049016793,0.0011111039],"category_scores_gemma":[0.051188923,0.00028997668,0.00055586576,0.00480855,0.00025241406,0.0010970305,0.00085467374,0.00061834,0.0003151967],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00170997,0.00032188484,0.44333068,0.00032758364,0.0005896595,0.00035042514,0.00080596743,0.08828029,0.002904171,0.004549549,0.008045609,0.4487843],"study_design_scores_gemma":[0.000168021,0.0012324789,0.24411012,0.00009837171,0.0002605899,0.0005543712,0.00028192575,0.7327342,0.008404432,0.0058714794,0.006012205,0.00027177032],"about_ca_topic_score_codex":0.0078100264,"about_ca_topic_score_gemma":0.005894295,"teacher_disagreement_score":0.0078100264,"about_ca_system_score_codex":0.00070128933,"about_ca_system_score_gemma":0.0011643154,"threshold_uncertainty_score":0.037592053},"labels":[],"label_agreement":null},{"id":"W2116887614","doi":"10.1093/biostatistics/kxq079","title":"A particular diffusion model for incomplete longitudinal data: application to the multicenter AIDS cohort study","year":2011,"lang":"en","type":"article","venue":"Biostatistics","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Missing data; Covariate; Bayesian probability; Gibbs sampling; Longitudinal data; Multicenter AIDS Cohort Study; Statistics; Sampling (signal processing); Computer science; Econometrics; Mathematics; Medicine; Data mining; Human immunodeficiency virus (HIV)","score_opus":0.09837403590722711,"score_gpt":0.33226683152782255,"score_spread":0.23389279562059545,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2116887614","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01855439,0.0010929253,0.9764713,0.002280531,0.00010246366,0.00014715125,0.00040128542,0.00016824844,0.0007817869],"genre_scores_gemma":[0.39686167,0.0043730834,0.5873361,0.0009888174,0.0005768007,0.0017649032,0.0017531981,0.00020681602,0.0061385427],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9902852,0.0074387807,0.0003032524,0.0009632232,0.0007260979,0.00028341028],"domain_scores_gemma":[0.94221085,0.050396692,0.0027004369,0.0023944925,0.0014893486,0.0008082257],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.033827607,0.0013078476,0.0030947865,0.0018844806,0.0015753335,0.0021101828,0.0047339173,0.0034306392,0.0033646566],"category_scores_gemma":[0.08051292,0.0011498109,0.0028398605,0.0031137175,0.0023155315,0.0035431464,0.0033617697,0.004585716,0.00051578286],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040877218,0.00015512844,0.012763908,0.0004195276,0.00086910644,0.0011967185,0.0012117295,0.2972086,0.000924666,0.62340456,0.0064681745,0.054969072],"study_design_scores_gemma":[0.00020882099,0.00007679868,0.0011904147,0.00007534896,0.00016579137,0.0004069603,0.00011163395,0.74943,0.00021973529,0.24424432,0.0038048937,0.000065385764],"about_ca_topic_score_codex":0.014346151,"about_ca_topic_score_gemma":0.0111192865,"teacher_disagreement_score":0.033827607,"about_ca_system_score_codex":0.0020016143,"about_ca_system_score_gemma":0.002427928,"threshold_uncertainty_score":0.17889959},"labels":[],"label_agreement":null},{"id":"W2117015589","doi":"10.1093/biostatistics/kxu011","title":"Composite likelihood for joint analysis of multiple multistate processes via copulas","year":2014,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Copula (linguistics); Inference; Computer science; Joint (building); Econometrics; Markov model; Markov chain; Machine learning; Mathematics; Artificial intelligence","score_opus":0.07492729461245784,"score_gpt":0.35729605603842446,"score_spread":0.2823687614259666,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2117015589","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010874153,0.00009665528,0.9983224,0.00006129756,0.000009626144,0.000016023794,0.00003065081,0.00005637751,0.0003194244],"genre_scores_gemma":[0.28351286,0.0021146752,0.70400375,0.0002904792,0.0002005556,0.0008420567,0.00064742606,0.0004732295,0.007915046],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9970823,0.0017006248,0.00011763499,0.00035630958,0.00054961146,0.00019351128],"domain_scores_gemma":[0.9859384,0.011245914,0.001009748,0.00082970894,0.0007024521,0.0002737602],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008303401,0.0012834918,0.0017145372,0.0017488694,0.000682538,0.0023209432,0.0019646746,0.0015791024,0.0040788394],"category_scores_gemma":[0.027382273,0.00094037515,0.0020978893,0.0018057722,0.0021476806,0.002765625,0.0033110862,0.0040510506,0.00091739994],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004566009,0.000054634216,0.001827836,0.00019817127,0.000117582014,0.00027662033,0.0002173052,0.39051902,0.001311055,0.57483566,0.0017244369,0.028872052],"study_design_scores_gemma":[0.0000061088103,0.000015357797,0.0002822301,0.000027540698,0.000018514018,0.000058144185,0.000021193173,0.85344994,0.00028268236,0.14451408,0.0013030886,0.000021134196],"about_ca_topic_score_codex":0.0032664742,"about_ca_topic_score_gemma":0.0024810384,"teacher_disagreement_score":0.008303401,"about_ca_system_score_codex":0.0013583525,"about_ca_system_score_gemma":0.0022491694,"threshold_uncertainty_score":0.043913126},"labels":[],"label_agreement":null},{"id":"W2121713677","doi":"10.1093/biostatistics/kxq067","title":"Functional mixture regression","year":2010,"lang":"en","type":"article","venue":"Biostatistics","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University; University of Toronto","funders":"","keywords":"Regression; Functional principal component analysis; Regression analysis; Computer science; Principal component analysis; Functional data analysis; Scalar (mathematics); Linear regression; Artificial intelligence; Machine learning; Econometrics; Mathematics; Statistics","score_opus":0.015400804457123302,"score_gpt":0.2671184447815994,"score_spread":0.2517176403244761,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121713677","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017979745,0.0003931165,0.9963888,0.00016364125,0.000049270933,0.00005290801,0.00015139596,0.00029795134,0.0007049792],"genre_scores_gemma":[0.1918047,0.0019879537,0.7918605,0.0005831597,0.00040203135,0.0010142146,0.0018897511,0.000724934,0.009732709],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9920859,0.0053711417,0.00025182354,0.001197624,0.0008383526,0.00025513626],"domain_scores_gemma":[0.98866105,0.008102775,0.0006766755,0.0014357376,0.0009828637,0.00014101039],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013744086,0.0026551965,0.0026632447,0.002701771,0.0009561946,0.0022966627,0.0038027433,0.0032476503,0.007734327],"category_scores_gemma":[0.035303116,0.0011403728,0.0039096186,0.0033296184,0.0015021712,0.0031251588,0.0027213385,0.0034833679,0.003710828],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037551054,0.00016991766,0.005415654,0.0006657014,0.001013942,0.00024122641,0.00039139713,0.30566368,0.0027501606,0.29661238,0.010897904,0.37580243],"study_design_scores_gemma":[0.000034961082,0.00008427722,0.0011740482,0.000091299466,0.0001332302,0.00018645581,0.000046943813,0.88397545,0.0008090644,0.10457212,0.008819068,0.000073122166],"about_ca_topic_score_codex":0.006427393,"about_ca_topic_score_gemma":0.005089974,"teacher_disagreement_score":0.013744086,"about_ca_system_score_codex":0.0012338074,"about_ca_system_score_gemma":0.0017626112,"threshold_uncertainty_score":0.07268655},"labels":[],"label_agreement":null},{"id":"W2123633327","doi":"10.1093/biostatistics/kxm025","title":"Regression analysis of multivariate panel count data","year":2007,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Multivariate statistics; Covariate; Statistics; Count data; Regression analysis; Econometrics; Regression; Marginal model; Event (particle physics); Multivariate analysis; Reliability (semiconductor); Mathematics","score_opus":0.3188325730028094,"score_gpt":0.47258056925801745,"score_spread":0.15374799625520802,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2123633327","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019928202,0.00067188364,0.97666204,0.00043082068,0.0000361122,0.00007377244,0.0010909742,0.0002768197,0.0008294669],"genre_scores_gemma":[0.64822507,0.0038763098,0.33691362,0.0003424495,0.00038130063,0.00084685243,0.0036826448,0.00018112891,0.005550678],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9953674,0.0034499834,0.00013003136,0.00049446,0.00039148293,0.00016672062],"domain_scores_gemma":[0.97817695,0.017744144,0.0020791956,0.0013763438,0.000481117,0.00014220529],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007054637,0.00088927976,0.0013968778,0.0013711988,0.00026488988,0.00094133255,0.0014221721,0.001019449,0.0043492857],"category_scores_gemma":[0.029495204,0.00031138526,0.0011400107,0.0025723039,0.00056619477,0.0009959181,0.0009954694,0.0019252478,0.0007788185],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021233,0.00024050081,0.03252966,0.00054244493,0.0007949742,0.0007500596,0.0003179169,0.580825,0.0033577564,0.1938018,0.005134015,0.18149352],"study_design_scores_gemma":[0.00002264929,0.00012165954,0.00807969,0.00004116229,0.00008622488,0.00013539422,0.00006322621,0.90070987,0.00094051886,0.08456633,0.005187699,0.00004554543],"about_ca_topic_score_codex":0.0032489826,"about_ca_topic_score_gemma":0.0019283107,"teacher_disagreement_score":0.007054637,"about_ca_system_score_codex":0.00040761256,"about_ca_system_score_gemma":0.000615892,"threshold_uncertainty_score":0.03730893},"labels":[],"label_agreement":null},{"id":"W2126930228","doi":"10.1093/biostatistics/4.3.365","title":"A closer look at the distribution of number needed to treat (NNT): a Bayesian approach","year":2003,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Number needed to treat; Point estimation; Posterior probability; Bayesian probability; Mathematics; Statistics; Prior probability; Econometrics; Computer science; Confidence interval; Relative risk","score_opus":0.28032836460518146,"score_gpt":0.4902424701533613,"score_spread":0.20991410554817985,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2126930228","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028073746,0.008323174,0.95359397,0.027781576,0.0005163788,0.00005856885,0.00013782497,0.00009757385,0.0066835633],"genre_scores_gemma":[0.2514544,0.03454903,0.6756006,0.019543892,0.006315019,0.0007947165,0.00051908905,0.000490508,0.010732787],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9843387,0.011463721,0.00038033054,0.0013223863,0.0022704098,0.00022442068],"domain_scores_gemma":[0.9323472,0.062094714,0.00143686,0.0016647461,0.0020572639,0.00039924044],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0225364,0.0009160641,0.002554928,0.0028571454,0.0006985079,0.0037640403,0.0021945501,0.0040969397,0.0064222305],"category_scores_gemma":[0.11450959,0.00068048656,0.0013105995,0.0020932218,0.0034864596,0.0077836514,0.0015679604,0.010682454,0.0011786674],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011633678,0.000085454834,0.0023164768,0.00050697644,0.0002267298,0.00022972774,0.0004786832,0.05487243,0.0006255898,0.8231188,0.011292315,0.10613041],"study_design_scores_gemma":[0.000056623005,0.00010669645,0.0011604981,0.00037783946,0.00005175194,0.0004652347,0.00013460786,0.086255684,0.0002716857,0.88819146,0.022851117,0.0000766993],"about_ca_topic_score_codex":0.0028345694,"about_ca_topic_score_gemma":0.0019619043,"teacher_disagreement_score":0.9774636,"about_ca_system_score_codex":0.0025496413,"about_ca_system_score_gemma":0.0016305752,"threshold_uncertainty_score":0.11918527},"labels":[],"label_agreement":null},{"id":"W2127195807","doi":"10.1093/biostatistics/4.3.479","title":"Conditional analysis of mixed Poisson processes with baseline counts: implications for trial design and analysis","year":2003,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Statistics; Baseline (sea); Sample size determination; Count data; Negative binomial distribution; Multiple baseline design; Poisson distribution; Randomization; Mathematics; Computer science; Econometrics; Clinical trial; Medicine; Internal medicine","score_opus":0.4690235198847609,"score_gpt":0.5372107865117601,"score_spread":0.06818726662699925,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127195807","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017321244,0.0011326863,0.99389184,0.0015739431,0.00015967422,0.00077185244,0.00011093797,0.00013215559,0.00049490883],"genre_scores_gemma":[0.05637375,0.002376362,0.9283614,0.0018141129,0.00043174683,0.009374337,0.00025943064,0.0001309258,0.0008778929],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.68340623,0.29106,0.0064218678,0.006117291,0.011809844,0.0011846509],"domain_scores_gemma":[0.36716637,0.59129477,0.017019618,0.017020958,0.006122682,0.0013756572],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.2938994,0.0024688495,0.006307442,0.0028122277,0.0011679211,0.0038703764,0.0046299,0.004473187,0.0067984043],"category_scores_gemma":[0.5427244,0.0016032989,0.0037798928,0.0052129673,0.0064732777,0.0066937483,0.004822844,0.007478864,0.001040492],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019550682,0.00028575267,0.005016419,0.0019711764,0.0016084759,0.00048371582,0.0006846879,0.053695433,0.0007804954,0.7823526,0.007104655,0.14406142],"study_design_scores_gemma":[0.0008398354,0.000729365,0.0014039228,0.00040654026,0.0003893303,0.00024924552,0.000063893225,0.19924657,0.0007316846,0.7892795,0.006565856,0.00009419143],"about_ca_topic_score_codex":0.0014820809,"about_ca_topic_score_gemma":0.0013097266,"teacher_disagreement_score":0.2938994,"about_ca_system_score_codex":0.0031628236,"about_ca_system_score_gemma":0.008692903,"threshold_uncertainty_score":0.8707478},"labels":[],"label_agreement":null},{"id":"W2128514707","doi":"10.1093/biostatistics/kxm029","title":"Joint inference for nonlinear mixed-effects models and time to event at the presence of missing data","year":2007,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health","keywords":"Covariate; Inference; Missing data; Event (particle physics); Computer science; Econometrics; Proportional hazards model; Mixed model; Process (computing); Statistical inference; Longitudinal data; Counting process; Event data; Statistics; Data mining; Artificial intelligence; Mathematics; Machine learning","score_opus":0.13065622976442884,"score_gpt":0.41384063250208264,"score_spread":0.2831844027376538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128514707","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043180953,0.0003645289,0.99414814,0.00050852704,0.000039709925,0.000041362968,0.000110450506,0.0000993818,0.00036983375],"genre_scores_gemma":[0.2453944,0.002150562,0.7459981,0.00056788453,0.0003932406,0.0010816155,0.000946606,0.00016596903,0.003301729],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9766816,0.018662682,0.0006525983,0.0023678197,0.0012833981,0.0003519361],"domain_scores_gemma":[0.8692023,0.12017671,0.0045844247,0.0038946823,0.001546921,0.000594946],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.046703573,0.0017588012,0.0030420606,0.0030935083,0.0014145593,0.002556992,0.004258338,0.0027006073,0.004510897],"category_scores_gemma":[0.16242544,0.0017779472,0.0031286757,0.0030422779,0.004237436,0.0050378176,0.0033672494,0.0043574288,0.00062546524],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024877704,0.00013813896,0.0092963185,0.00050961925,0.0009873325,0.0006431381,0.0009013179,0.12370744,0.00047979466,0.80171126,0.001857472,0.059519533],"study_design_scores_gemma":[0.00008925718,0.000062840656,0.0009924211,0.00008508017,0.00015373448,0.00016143796,0.00008723628,0.3402221,0.00026294673,0.6557756,0.0020686043,0.000038729777],"about_ca_topic_score_codex":0.009327344,"about_ca_topic_score_gemma":0.0076434934,"teacher_disagreement_score":0.046703573,"about_ca_system_score_codex":0.002465844,"about_ca_system_score_gemma":0.0038152442,"threshold_uncertainty_score":0.24699509},"labels":[],"label_agreement":null},{"id":"W2128581110","doi":"10.1093/biostatistics/3.3.299","title":"Experimental design issues for the early detection of disease: novel designs","year":2002,"lang":"en","type":"article","venue":"Biostatistics","topic":"Global Cancer Incidence and Screening","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Cancer Institute; National Institutes of Health","keywords":"Medicine; Family medicine; Group (periodic table); Outcome (game theory); Medical physics; Mathematics","score_opus":0.23445914822665576,"score_gpt":0.37581462923286985,"score_spread":0.1413554810062141,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128581110","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025988942,0.0029682538,0.9373351,0.0041995677,0.003074222,0.020549566,0.00024427587,0.0002851716,0.005354921],"genre_scores_gemma":[0.15676372,0.0013059797,0.7516757,0.0030124697,0.0018073276,0.08379001,0.0001145315,0.00009201698,0.0014382391],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.53217936,0.4263576,0.008957294,0.013620402,0.017317442,0.0015679126],"domain_scores_gemma":[0.32208288,0.6219121,0.017342543,0.03005505,0.007516246,0.0010911835],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3349799,0.0028005315,0.004984523,0.0017339225,0.0019444885,0.0048686783,0.0037174076,0.007035117,0.00744833],"category_scores_gemma":[0.5082751,0.0018866712,0.0038817152,0.0015615321,0.008962592,0.0072542634,0.003365336,0.0061350935,0.00071317045],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.02825685,0.0033444634,0.006557887,0.0096276915,0.005991836,0.0006256523,0.003823823,0.018916178,0.0035031429,0.67297596,0.006008786,0.24036774],"study_design_scores_gemma":[0.033378474,0.037899766,0.008248011,0.0032202236,0.0038083747,0.00067081786,0.00064072484,0.12416876,0.0058959927,0.7435599,0.037887152,0.00062179915],"about_ca_topic_score_codex":0.0004255295,"about_ca_topic_score_gemma":0.00036360932,"teacher_disagreement_score":0.6650201,"about_ca_system_score_codex":0.003476965,"about_ca_system_score_gemma":0.0034517255,"threshold_uncertainty_score":0.8200882},"labels":[],"label_agreement":null},{"id":"W2134647083","doi":"10.1093/biostatistics/kxn037","title":"Statistical monitoring of clinical trials with multivariate response and/or multiple arms: a flexible approach","year":2008,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Institute of Allergy and Infectious Diseases; ACT Government","keywords":"Interim; Multivariate statistics; Clinical trial; Robustness (evolution); Computer science; Randomized controlled trial; Interim analysis; Multivariate analysis; Statistics; Medicine; Mathematics; Machine learning; Internal medicine","score_opus":0.8559835100558566,"score_gpt":0.644237374207395,"score_spread":0.21174613584846158,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134647083","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005780034,0.00041902403,0.9917983,0.0008157484,0.000048843925,0.0004219741,0.000056933426,0.00020083251,0.0004583247],"genre_scores_gemma":[0.24913096,0.0006617624,0.7460454,0.0006870457,0.00019849774,0.0023073812,0.00013581397,0.00015689772,0.00067618437],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.73107785,0.24006599,0.0064089345,0.00739003,0.013724834,0.0013324114],"domain_scores_gemma":[0.4360677,0.4747637,0.03228054,0.048623003,0.0065571885,0.0017079228],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.23669976,0.0021604956,0.0039316746,0.0033310477,0.0010622876,0.0041391007,0.005066133,0.0033671681,0.0028664826],"category_scores_gemma":[0.46518403,0.001628162,0.004092996,0.003428879,0.0060501797,0.0065105427,0.006384916,0.008714853,0.00047052288],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00218739,0.00025909021,0.007425096,0.00091125397,0.002646805,0.00052646967,0.0011948656,0.42568737,0.0017999174,0.29089153,0.0025392578,0.26393098],"study_design_scores_gemma":[0.000759845,0.0006526631,0.0019259624,0.0002466512,0.000380192,0.00016028562,0.00006748402,0.6346979,0.001361209,0.3567281,0.0029015855,0.00011817503],"about_ca_topic_score_codex":0.0015980806,"about_ca_topic_score_gemma":0.0012115932,"teacher_disagreement_score":0.23669976,"about_ca_system_score_codex":0.0027880112,"about_ca_system_score_gemma":0.0041780104,"threshold_uncertainty_score":0.9412851},"labels":[],"label_agreement":null},{"id":"W2135189416","doi":"10.1093/biostatistics/kxg041","title":"Analysis of longitudinal marginal structural models","year":2004,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Estimator; Censoring (clinical trials); Inverse probability; Marginal structural model; Confounding; Statistics; Econometrics; Mathematics; Computer science","score_opus":0.15749982260053155,"score_gpt":0.4178060521425315,"score_spread":0.260306229542,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135189416","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027388493,0.0005774605,0.96956664,0.00076006656,0.00003815985,0.00008028994,0.00039471764,0.00023046677,0.00096376595],"genre_scores_gemma":[0.69962096,0.0018379148,0.29018918,0.00046476698,0.0002722264,0.0010764259,0.0016955596,0.00016681576,0.0046761683],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9922415,0.005923012,0.00018036354,0.00070870406,0.00065944873,0.00028696287],"domain_scores_gemma":[0.94635755,0.045303416,0.003154156,0.0029985162,0.0016624586,0.0005239074],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018808352,0.0007913011,0.0015270755,0.002004716,0.00057762273,0.0015242796,0.002419782,0.0012454191,0.008210074],"category_scores_gemma":[0.06943927,0.00069830095,0.0020458086,0.0016075645,0.0015512797,0.002596901,0.002481475,0.0019825085,0.0005705807],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015772728,0.00013626162,0.015399318,0.00032059968,0.00066459196,0.00018684298,0.0005958049,0.15537402,0.0004303099,0.77573746,0.0023722614,0.04862488],"study_design_scores_gemma":[0.000048931997,0.0000796866,0.0019961067,0.000050089904,0.00008832329,0.000057480298,0.000085909945,0.5435757,0.00023489802,0.45158157,0.0021746447,0.000026639256],"about_ca_topic_score_codex":0.0036736117,"about_ca_topic_score_gemma":0.0032533247,"teacher_disagreement_score":0.018808352,"about_ca_system_score_codex":0.0015211651,"about_ca_system_score_gemma":0.001908118,"threshold_uncertainty_score":0.099469244},"labels":[],"label_agreement":null},{"id":"W2138916823","doi":"10.1093/biostatistics/kxs024","title":"Targeted maximum likelihood estimation for marginal time-dependent treatment effects under density misspecification","year":2012,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University Health Centre","funders":"Canadian Institutes of Health Research; McGill University","keywords":"Estimator; Statistics; Generalized estimating equation; Mathematics; Econometrics; Delta method; Parametric statistics; Confounding; Random effects model; Variance (accounting); Computer science; Medicine; Meta-analysis","score_opus":0.08074750171030778,"score_gpt":0.3757055195828782,"score_spread":0.29495801787257037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2138916823","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004371953,0.00023587944,0.99478215,0.00015803745,0.000012647064,0.00005165332,0.000049516722,0.00011599949,0.00022214322],"genre_scores_gemma":[0.37687644,0.0011217378,0.617166,0.0005046583,0.0001281063,0.0011900875,0.0005531622,0.00018317,0.0022766145],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9860409,0.011603523,0.00035898827,0.00087412994,0.0008763695,0.0002460009],"domain_scores_gemma":[0.92484474,0.06730395,0.0028625904,0.0036266188,0.0011086303,0.00025342128],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.027295807,0.0011570426,0.0022601997,0.0015251653,0.0004964073,0.0015519084,0.0028042027,0.0019040684,0.0028269175],"category_scores_gemma":[0.118610606,0.00074963755,0.0021311815,0.0018617184,0.0021314714,0.0020591707,0.002552632,0.0028127867,0.00053848553],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062862464,0.00019624754,0.013974779,0.0010768626,0.0015242727,0.0006218575,0.0010256711,0.3440421,0.002562787,0.35089132,0.003404771,0.28005078],"study_design_scores_gemma":[0.0001544091,0.00017563297,0.0025499102,0.00011913778,0.00019801297,0.00022132939,0.00006501923,0.71136564,0.0014357424,0.281445,0.0022317092,0.00003839202],"about_ca_topic_score_codex":0.0016610197,"about_ca_topic_score_gemma":0.0014163285,"teacher_disagreement_score":0.027295807,"about_ca_system_score_codex":0.0010366505,"about_ca_system_score_gemma":0.002030303,"threshold_uncertainty_score":0.14435577},"labels":[],"label_agreement":null},{"id":"W2146901830","doi":"10.1093/biostatistics/kxu026","title":"Multiple comparison procedures for neuroimaging genomewide association studies","year":2014,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute on Aging; University of California, San Diego; Canadian Institutes of Health Research; University of California, Los Angeles; National Institutes of Health; Genentech; IXICO; Servier; Eisai; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; Alzheimer's Association; Amorfix Life Sciences; F. Hoffmann-La Roche; Medpace; AstraZeneca; Eli Lilly and Company; Bristol-Myers Squibb; Novartis Pharmaceuticals Corporation; Wellcome Trust; Synarc; Bayer HealthCare; Alzheimer's Disease Neuroimaging Initiative; National Center for Advancing Translational Sciences; Meso Scale Diagnostics; National Science Foundation","keywords":"Neuroimaging; Imaging genetics; Univariate; False discovery rate; Computer science; Alzheimer's Disease Neuroimaging Initiative; Multiple comparisons problem; Dimensionality reduction; Genome-wide association study; Brain size; Statistical power; Computational biology; Artificial intelligence; Machine learning; Single-nucleotide polymorphism; Neuroscience; Magnetic resonance imaging; Biology; Multivariate statistics; Medicine; Statistics; Genetics; Gene; Genotype; Mathematics; Cognition","score_opus":0.03238930918415187,"score_gpt":0.3310335732112797,"score_spread":0.2986442640271278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146901830","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009578026,0.00041232014,0.9961392,0.00030780133,0.00030468366,0.0005418923,0.00028185136,0.00056545774,0.00048893836],"genre_scores_gemma":[0.018973392,0.00042646666,0.97203773,0.00026942423,0.00031962682,0.0063906256,0.00046212462,0.00036055795,0.00076005067],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.81071293,0.16661249,0.0049641225,0.0065695113,0.010195052,0.0009459062],"domain_scores_gemma":[0.69136167,0.27528948,0.006322824,0.020576889,0.0055807983,0.00086834043],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.121369846,0.0032407378,0.004492872,0.0052787843,0.0020241647,0.002243679,0.0065418924,0.003991575,0.018117866],"category_scores_gemma":[0.3328558,0.0015513151,0.0044824057,0.007793217,0.004132033,0.003122651,0.003951554,0.0106240595,0.0034869965],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00243203,0.0007523401,0.004925753,0.0027537544,0.0056997226,0.0014916363,0.0011805813,0.038145375,0.0040918435,0.40080413,0.027842091,0.50988066],"study_design_scores_gemma":[0.0010795647,0.0020317682,0.005093478,0.00054779137,0.00079429743,0.0011351806,0.00031460822,0.26896214,0.0045073717,0.6632606,0.052000973,0.00027227256],"about_ca_topic_score_codex":0.0015792158,"about_ca_topic_score_gemma":0.0011239868,"teacher_disagreement_score":0.121369846,"about_ca_system_score_codex":0.0016196604,"about_ca_system_score_gemma":0.0041694324,"threshold_uncertainty_score":0.64187276},"labels":[],"label_agreement":null},{"id":"W2148628375","doi":"10.1093/biostatistics/4.2.167","title":"Analysis of a molecular genetic neuro-oncology study with partially biased selection","year":2003,"lang":"en","type":"article","venue":"Biostatistics","topic":"Glioma Diagnosis and Treatment","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; Ottawa Regional Cancer Foundation","funders":"National Cancer Institute; National Institutes of Health","keywords":"Medicine; Chemotherapy; Oncology; Internal medicine; Disease; Radiation therapy; Population; Cancer; Radiation oncology","score_opus":0.01655065745641814,"score_gpt":0.29282907121953483,"score_spread":0.2762784137631167,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2148628375","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9982578,0.00012257777,0.0009886932,0.00018679205,0.000006364799,0.000050040588,0.0000964243,0.000005346553,0.00028599263],"genre_scores_gemma":[0.9981501,0.00004314726,0.0013325126,0.00014754458,0.00001591771,0.000058741472,0.00014233921,0.0000039191377,0.000105745756],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9680324,0.027413145,0.0009577549,0.0012983948,0.0014553819,0.0008428303],"domain_scores_gemma":[0.9718354,0.017340912,0.0057842354,0.003451511,0.00080655294,0.000781388],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.019412093,0.00066402793,0.00081136596,0.0014535824,0.0013264247,0.0011371596,0.0010959713,0.0011558738,0.0010353847],"category_scores_gemma":[0.039649304,0.0004199138,0.001354156,0.002048821,0.00119301,0.00033554967,0.0010347038,0.0006867514,0.00020620452],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087458786,0.00011058973,0.9910678,0.000023527195,0.0006106635,0.0010792903,0.0005038974,0.00037416557,0.0010117035,0.00047150877,0.00020202129,0.0036701974],"study_design_scores_gemma":[0.00046851428,0.0026362604,0.97606885,0.000034143846,0.0013141863,0.0049697724,0.00086352177,0.009167149,0.0015000908,0.0011192545,0.001823205,0.000035148354],"about_ca_topic_score_codex":0.0037510828,"about_ca_topic_score_gemma":0.0051015024,"teacher_disagreement_score":0.9805879,"about_ca_system_score_codex":0.00072739995,"about_ca_system_score_gemma":0.0011826031,"threshold_uncertainty_score":0.102662146},"labels":[],"label_agreement":null},{"id":"W2153510570","doi":"10.1093/biostatistics/kxq048","title":"Feature selection in finite mixture of sparse normal linear models in high-dimensional feature space","year":2010,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; McGill University","funders":"","keywords":"Feature selection; Curse of dimensionality; Computer science; Bayesian information criterion; Maximization; Feature vector; Feature (linguistics); Boosting (machine learning); Bayesian probability; Data mining; Machine learning; Algorithm; Artificial intelligence; Mathematical optimization; Mathematics","score_opus":0.03465438392887245,"score_gpt":0.3177505091851537,"score_spread":0.28309612525628125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153510570","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008787558,0.0001061312,0.9907284,0.00009343385,0.000007321719,0.000017155911,0.000022771605,0.00013992868,0.00009732526],"genre_scores_gemma":[0.44046342,0.0004244847,0.5559343,0.00021781526,0.00013176724,0.0004745139,0.0005162464,0.00009405923,0.0017433616],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99680483,0.002170758,0.00009716951,0.00035898742,0.00043430296,0.00013389802],"domain_scores_gemma":[0.99351805,0.0051393346,0.00043957884,0.00037530175,0.00041659034,0.00011106072],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0056029097,0.0006034295,0.0014952865,0.00084453536,0.00051907153,0.0009098892,0.0012640427,0.0010158007,0.000712825],"category_scores_gemma":[0.0121966945,0.000664782,0.0012571287,0.0010862439,0.00113447,0.0011379714,0.0012234616,0.0011833586,0.0003799666],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031452926,0.0001097981,0.0029426415,0.00016354503,0.00014960792,0.00019422304,0.0002473036,0.79770285,0.0037160993,0.038394492,0.0015380434,0.15452689],"study_design_scores_gemma":[0.00001334148,0.000025723119,0.00027171895,0.0000040760483,0.0000066210405,0.000019842608,0.0000054849174,0.9859563,0.0002917847,0.013179304,0.00021725433,0.0000085615],"about_ca_topic_score_codex":0.002572561,"about_ca_topic_score_gemma":0.0023633577,"teacher_disagreement_score":0.0056029097,"about_ca_system_score_codex":0.00071450486,"about_ca_system_score_gemma":0.00073138735,"threshold_uncertainty_score":0.029631376},"labels":[],"label_agreement":null},{"id":"W2156745572","doi":"10.1093/biostatistics/kxp041","title":"Bayesian random-effects threshold regression with application to survival data with nonproportional hazards","year":2009,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Distribution Estimation and Applications","field":"Mathematics","cited_by":33,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Institute for Occupational Safety and Health","keywords":"Bayesian probability; Regression; Statistics; Proportional hazards model; Regression analysis; Econometrics; Computer science; Mathematics","score_opus":0.03665299856253803,"score_gpt":0.35921251095346285,"score_spread":0.32255951239092484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156745572","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004104533,0.00028369913,0.99489814,0.00016625994,0.000017394877,0.000050427436,0.00003952109,0.0001958735,0.00024409677],"genre_scores_gemma":[0.24350259,0.0011943185,0.7516418,0.00024700348,0.00009520526,0.00069780607,0.0003282848,0.00021881291,0.0020741886],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9925875,0.0054029715,0.0003062181,0.0007148394,0.00078450737,0.00020398738],"domain_scores_gemma":[0.9650938,0.030241465,0.0016217652,0.0017125333,0.0010592735,0.00027117826],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.020017631,0.00087461097,0.0018307352,0.0018475588,0.0006808547,0.0012593481,0.0023824729,0.0019361814,0.0024383988],"category_scores_gemma":[0.073184155,0.0008981687,0.0017534948,0.0025030524,0.0015469268,0.0019696006,0.0021816595,0.0029157873,0.000589531],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029981314,0.0001475583,0.0061173495,0.00038957188,0.0003830091,0.00051247,0.00048089196,0.5532603,0.0016887296,0.26938802,0.002015333,0.16531692],"study_design_scores_gemma":[0.00007127548,0.00005900233,0.000927572,0.000047401885,0.000044142897,0.00015731223,0.00002813039,0.8692432,0.00052032917,0.12742075,0.0014471284,0.000033701173],"about_ca_topic_score_codex":0.008354096,"about_ca_topic_score_gemma":0.005639373,"teacher_disagreement_score":0.020017631,"about_ca_system_score_codex":0.0014649607,"about_ca_system_score_gemma":0.0024482263,"threshold_uncertainty_score":0.105864644},"labels":[],"label_agreement":null},{"id":"W2159353629","doi":"10.1093/biostatistics/5.1.129","title":"Tests for multivariate recurrent events in the presence of a terminal event","year":2004,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Univariate; Multivariate statistics; Medicine; Event (particle physics); Inference; Statistical hypothesis testing; Statistics; Breast cancer; Multivariate analysis; Econometrics; Cancer; Computer science; Internal medicine; Mathematics; Artificial intelligence","score_opus":0.5050057957657591,"score_gpt":0.5862204328749212,"score_spread":0.0812146371091621,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2159353629","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.75384533,0.0022240249,0.22364879,0.001765837,0.0005927309,0.00149524,0.0067012035,0.0011680505,0.008558673],"genre_scores_gemma":[0.9656913,0.00022932811,0.029246245,0.00032242242,0.00022278429,0.001190243,0.0020263917,0.00010307215,0.0009682133],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9301076,0.043252837,0.0047630267,0.008886352,0.009952461,0.0030376052],"domain_scores_gemma":[0.48273614,0.45804384,0.030683987,0.020310143,0.004806085,0.003419788],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.07715519,0.0016625173,0.0029540423,0.0042547206,0.0010014676,0.0024443227,0.004018327,0.0025121241,0.012357923],"category_scores_gemma":[0.27993643,0.0005111909,0.0051387455,0.0045884307,0.0033547112,0.004423012,0.0033968547,0.0047749775,0.0011753511],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.014843834,0.0012832247,0.7916312,0.0014518359,0.010072113,0.0025541813,0.001768178,0.008734103,0.0042116935,0.016928097,0.0045315973,0.14198989],"study_design_scores_gemma":[0.0016259609,0.021102007,0.758228,0.0004885716,0.0060677957,0.004515688,0.0024587794,0.12766588,0.007544292,0.057749152,0.012030345,0.0005235101],"about_ca_topic_score_codex":0.0012136677,"about_ca_topic_score_gemma":0.00093288044,"teacher_disagreement_score":0.07715519,"about_ca_system_score_codex":0.00078228355,"about_ca_system_score_gemma":0.0017702492,"threshold_uncertainty_score":0.40804052},"labels":[],"label_agreement":null},{"id":"W2163180646","doi":"10.1093/biostatistics/kxm053","title":"Efficient p-value estimation in massively parallel testing problems","year":2008,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; University of Toronto; SickKids Foundation; Hospital for Sick Children; Public Health Ontario; Toronto Public Health","funders":"National Cancer Institute","keywords":"Computer science; Bayesian probability; Test statistic; Statistic; Null (SQL); Set (abstract data type); Massively parallel; Fraction (chemistry); Null distribution; Permutation (music); Algorithm; Statistical hypothesis testing; Mathematics; Statistics; Data mining; Artificial intelligence; Parallel computing","score_opus":0.02946165175741932,"score_gpt":0.2710030859782423,"score_spread":0.24154143422082297,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163180646","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034025323,0.00013379102,0.9949032,0.000239879,0.000035731387,0.000077563855,0.000058237867,0.00066269044,0.00048640047],"genre_scores_gemma":[0.06605116,0.00013237711,0.9315254,0.00013418338,0.00012480625,0.0006136235,0.00029532865,0.0001935723,0.00092948636],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9925356,0.004552765,0.00042148752,0.0006808445,0.0015469644,0.00026244938],"domain_scores_gemma":[0.945684,0.04583551,0.0017914864,0.0037911085,0.0023161967,0.00058166153],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012932933,0.0017405569,0.002646478,0.0022297748,0.0015047108,0.002396779,0.0035487537,0.0015057657,0.004511343],"category_scores_gemma":[0.070148736,0.00123317,0.0015334907,0.0028665685,0.0024353168,0.0026087028,0.0038094295,0.0032163037,0.0017255751],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005442209,0.0002229273,0.0049623516,0.00041976757,0.00029803644,0.00089592946,0.0003699687,0.5414614,0.0036085153,0.084534,0.008767496,0.3539154],"study_design_scores_gemma":[0.00014474979,0.000063750405,0.00046139245,0.00002556097,0.000025218476,0.00019360521,0.000044330263,0.83441883,0.0012880223,0.16092788,0.002380462,0.000026117545],"about_ca_topic_score_codex":0.0035485278,"about_ca_topic_score_gemma":0.0044036107,"teacher_disagreement_score":0.012932933,"about_ca_system_score_codex":0.001276505,"about_ca_system_score_gemma":0.0032065806,"threshold_uncertainty_score":0.06839669},"labels":[],"label_agreement":null},{"id":"W2163628937","doi":"10.1093/biostatistics/kxn044","title":"Bias in 2-part mixed models for longitudinal semicontinuous data","year":2009,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":111,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"University of Toronto","keywords":"Random effects model; Mixed model; Censoring (clinical trials); Statistics; Missing data; Econometrics; Monte Carlo method; Independence (probability theory); Mathematics; Variance (accounting); Computer science; Medicine; Meta-analysis","score_opus":0.45668331579765237,"score_gpt":0.4407942544428088,"score_spread":0.01588906135484358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163628937","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010641432,0.00048865535,0.98767227,0.00044523113,0.000061147824,0.00011156772,0.00011467308,0.00019039457,0.00027459915],"genre_scores_gemma":[0.3402636,0.0011915147,0.65174943,0.0011963333,0.00024918595,0.0014441115,0.0007666952,0.00022747788,0.0029115987],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9422336,0.047476593,0.001997618,0.0042707515,0.0032237845,0.0007975874],"domain_scores_gemma":[0.6679393,0.30187768,0.011612894,0.014162258,0.003824858,0.00058308995],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.13979617,0.0016333876,0.0032303405,0.0027103904,0.0012438879,0.003560556,0.004547145,0.003828484,0.0027022902],"category_scores_gemma":[0.29047814,0.0017099605,0.0034834987,0.0030913593,0.005233283,0.0049346825,0.003583633,0.00410327,0.0006338868],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00086521596,0.00013011663,0.028644854,0.0009769605,0.0011767829,0.0013791181,0.0023898243,0.2507771,0.0016637887,0.6084085,0.0022282943,0.10135952],"study_design_scores_gemma":[0.00008527528,0.00018793771,0.0034984522,0.00020624115,0.00016688829,0.00044496878,0.00020157502,0.62323916,0.00093668,0.3686312,0.0023007107,0.00010084141],"about_ca_topic_score_codex":0.004003924,"about_ca_topic_score_gemma":0.004165641,"teacher_disagreement_score":0.13979617,"about_ca_system_score_codex":0.0025702703,"about_ca_system_score_gemma":0.0020876245,"threshold_uncertainty_score":0.73932165},"labels":[],"label_agreement":null},{"id":"W2163976725","doi":"10.1093/biostatistics/3.3.407","title":"A generalized mover-stayer model for panel data","year":2002,"lang":"en","type":"article","venue":"Biostatistics","topic":"Air Quality and Health Impacts","field":"Environmental Science","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Heart, Lung, and Blood Institute; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Covariate; Markov model; Markov chain; Markov process; Mathematics; Applied mathematics; Markov property; Computer science; Maximum likelihood; Statistics; Econometrics","score_opus":0.31346745460044456,"score_gpt":0.3641039722663774,"score_spread":0.050636517665932856,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2163976725","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0046010995,0.0004456571,0.98942554,0.0007050825,0.00012702873,0.00026727788,0.0019853099,0.00048713153,0.0019558812],"genre_scores_gemma":[0.22672085,0.0029976927,0.6935337,0.0011030667,0.0006639639,0.0043402375,0.011503244,0.0007372316,0.05839998],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9935135,0.003827268,0.00025937607,0.0011936922,0.0007271787,0.00047896797],"domain_scores_gemma":[0.9904198,0.0063786963,0.0009991247,0.0012348603,0.0007090516,0.0002585674],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012966352,0.0015483623,0.0024668586,0.0021794124,0.0012410418,0.0028781989,0.0071223998,0.0033687188,0.021716893],"category_scores_gemma":[0.02125062,0.0010509979,0.0030019959,0.0046557616,0.0016557967,0.0034408905,0.0024819968,0.0050047105,0.0058976123],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009365719,0.000097344615,0.0044148564,0.00021227555,0.00032571913,0.00057726173,0.0004147597,0.2364296,0.0007187503,0.70091045,0.014142304,0.041662984],"study_design_scores_gemma":[0.00006351208,0.00011585404,0.001721826,0.0000582684,0.00011521696,0.0003595927,0.00009283441,0.5989156,0.00018116669,0.3772316,0.02106036,0.00008424711],"about_ca_topic_score_codex":0.01576304,"about_ca_topic_score_gemma":0.014235688,"teacher_disagreement_score":0.021716893,"about_ca_system_score_codex":0.001962342,"about_ca_system_score_gemma":0.0027206165,"threshold_uncertainty_score":0.072650254},"labels":[],"label_agreement":null},{"id":"W2165213086","doi":"10.1093/biostatistics/kxj007","title":"Hybrid hierarchical clustering with applications to microarray data","year":2005,"lang":"en","type":"article","venue":"Biostatistics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":161,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Acadia University","funders":"National Cancer Institute; Natural Sciences and Engineering Research Council of Canada; Mitacs; National Institutes of Health; National Science Foundation","keywords":"Cluster analysis; Computer science; Hierarchical clustering; Data mining; Single-linkage clustering; Correlation clustering; Identification (biology); Cluster (spacecraft); CURE data clustering algorithm; Consensus clustering; Brown clustering; Artificial intelligence","score_opus":0.02578921853755773,"score_gpt":0.29945617986579215,"score_spread":0.2736669613282344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165213086","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030990103,0.0002440818,0.9954798,0.00011125391,0.000023646608,0.000035244095,0.000046353744,0.00059124304,0.00036926832],"genre_scores_gemma":[0.087074175,0.00030612765,0.9109157,0.00010449714,0.000089145855,0.00028555034,0.0002523253,0.00013944511,0.00083303853],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.995729,0.0022679202,0.00016783715,0.0005065952,0.0012059085,0.00012266447],"domain_scores_gemma":[0.99401456,0.0041816235,0.00023313238,0.00065439043,0.00079844374,0.00011788148],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0045452463,0.0008710457,0.0011390244,0.0030594901,0.0010960558,0.0010999873,0.0014883371,0.0010565454,0.0015825455],"category_scores_gemma":[0.011499013,0.00061804947,0.0013264789,0.004206644,0.0008310542,0.0009344657,0.0022197561,0.0012301966,0.00086284656],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020283496,0.00010550039,0.0030134784,0.00038456672,0.00039101637,0.00026223945,0.0004966881,0.44855213,0.012280855,0.046382427,0.004324565,0.48360375],"study_design_scores_gemma":[0.000016573626,0.000034477238,0.0007414517,0.000013929479,0.000026528334,0.00007312139,0.0000449707,0.9559575,0.0029214304,0.036434285,0.003704442,0.000031356933],"about_ca_topic_score_codex":0.0035015165,"about_ca_topic_score_gemma":0.0042407895,"teacher_disagreement_score":0.0045452463,"about_ca_system_score_codex":0.0007776546,"about_ca_system_score_gemma":0.00080979255,"threshold_uncertainty_score":0.024037838},"labels":[],"label_agreement":null},{"id":"W2166050184","doi":"10.1093/biostatistics/kxq041","title":"A multistate model for events defined by prolonged observation","year":2010,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of Toronto","keywords":"Duration (music); Event (particle physics); Spontaneous remission; Complete remission; State (computer science); Medicine; Computer science; Statistics; Mathematics; Internal medicine; Algorithm; Physics; Alternative medicine","score_opus":0.5493913029206225,"score_gpt":0.544768454180002,"score_spread":0.004622848740620578,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2166050184","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01857888,0.00043352295,0.9740292,0.0017574466,0.00017528058,0.00023592947,0.0015690839,0.00028006258,0.0029407365],"genre_scores_gemma":[0.7073056,0.001417751,0.24892448,0.0010233325,0.0005365895,0.0035717322,0.0030973735,0.00015544693,0.033967696],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99587005,0.002058874,0.00019982702,0.001069471,0.00040149537,0.0004003365],"domain_scores_gemma":[0.9833906,0.013301606,0.0013431276,0.00095218024,0.0006543104,0.00035830995],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010793244,0.0014963363,0.0020666302,0.0013920837,0.0007793575,0.0028208913,0.0036887694,0.0036554968,0.012067585],"category_scores_gemma":[0.016893627,0.00081031857,0.0023941544,0.0018220588,0.0021255042,0.004104357,0.0024010818,0.0054762745,0.001760587],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044530453,0.00013509199,0.004904969,0.00021595752,0.0002001069,0.0003921757,0.00057862105,0.46957842,0.00088759896,0.50337327,0.0025986254,0.016689846],"study_design_scores_gemma":[0.00007336569,0.00014785973,0.0007073819,0.00003610654,0.00007242126,0.00011671328,0.000065390785,0.85918444,0.0002134645,0.13617137,0.0031638085,0.000047724356],"about_ca_topic_score_codex":0.007685698,"about_ca_topic_score_gemma":0.0073794248,"teacher_disagreement_score":0.012067585,"about_ca_system_score_codex":0.0019419388,"about_ca_system_score_gemma":0.0024557249,"threshold_uncertainty_score":0.057080805},"labels":[],"label_agreement":null},{"id":"W2168126634","doi":"10.1093/biostatistics/kxs028","title":"Testing multiple variance components in linear mixed-effects models","year":2012,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Mathematics; Statistics; Test statistic; Null distribution; Estimator; Variance (accounting); Statistical hypothesis testing; Variance-based sensitivity analysis; One-way analysis of variance; Context (archaeology); Analysis of variance","score_opus":0.17829970945503099,"score_gpt":0.36776649057259386,"score_spread":0.18946678111756288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2168126634","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012332291,0.00044123194,0.98586535,0.0003430524,0.000076329525,0.00017614312,0.00012677001,0.00023919552,0.00039962053],"genre_scores_gemma":[0.2936103,0.00085367047,0.70157206,0.0004033473,0.00022150793,0.0019359646,0.00050754426,0.00016881595,0.0007267484],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.87572604,0.1085251,0.003123088,0.006652115,0.005241186,0.00073246745],"domain_scores_gemma":[0.6706559,0.3139336,0.005418079,0.0069587994,0.0025503992,0.0004831409],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0712146,0.002110912,0.0035250196,0.003275611,0.0013038571,0.0031324166,0.0035817898,0.0033146557,0.0035246727],"category_scores_gemma":[0.24208568,0.0012747573,0.0041902727,0.0033740797,0.0041395766,0.003040592,0.0031754572,0.0035354446,0.000556425],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00200984,0.0005518119,0.05064761,0.0030583316,0.007380927,0.0017884949,0.0021218779,0.15573122,0.0038479208,0.34512445,0.003571144,0.42416644],"study_design_scores_gemma":[0.00034967324,0.0010553253,0.008988221,0.00039160126,0.00085972936,0.0005009492,0.0003449037,0.46510977,0.0031880557,0.515569,0.0034387684,0.0002039942],"about_ca_topic_score_codex":0.0019806826,"about_ca_topic_score_gemma":0.0017094845,"teacher_disagreement_score":0.0712146,"about_ca_system_score_codex":0.001140539,"about_ca_system_score_gemma":0.0024818492,"threshold_uncertainty_score":0.37662333},"labels":[],"label_agreement":null},{"id":"W2170016983","doi":"10.1093/biostatistics/kxm054","title":"A simulation-based marginal method for longitudinal data with dropout and mismeasured covariates","year":2008,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Heart, Lung, and Blood Institute","keywords":"Covariate; Missing data; Estimator; Observational error; Statistics; Computer science; Econometrics; Dropout (neural networks); Framingham Heart Study; Data set; Inference; Mathematics; Artificial intelligence; Machine learning","score_opus":0.18878984038607685,"score_gpt":0.42635502145611714,"score_spread":0.2375651810700403,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170016983","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00063042965,0.0000930977,0.9988136,0.00008190043,0.000020849651,0.00004097453,0.0000282453,0.00011061499,0.00018031236],"genre_scores_gemma":[0.06857351,0.0006542673,0.9265342,0.00021552156,0.0001532502,0.0010087965,0.00037527992,0.00021224194,0.002272933],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99415994,0.0046777674,0.00013794434,0.00040074877,0.0005237775,0.00009975978],"domain_scores_gemma":[0.9816066,0.014845171,0.00067266234,0.0014336089,0.0010920404,0.00034999388],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0152488155,0.0008880935,0.0016418002,0.0020924879,0.00082368276,0.0011531695,0.0033432848,0.0012980989,0.00720063],"category_scores_gemma":[0.04656116,0.0008074898,0.001963782,0.0018927185,0.0017789953,0.0021588483,0.002607444,0.0028233458,0.0012568077],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028310574,0.00015560165,0.0043838494,0.00038375173,0.00048470998,0.00026847658,0.000478392,0.3177242,0.001300318,0.46842295,0.0056600906,0.20045461],"study_design_scores_gemma":[0.00007898371,0.00008011314,0.00055015786,0.00006766105,0.000058104953,0.00013899707,0.000029919396,0.85754746,0.00036242494,0.13461256,0.006429839,0.00004373987],"about_ca_topic_score_codex":0.0043039587,"about_ca_topic_score_gemma":0.004867422,"teacher_disagreement_score":0.0152488155,"about_ca_system_score_codex":0.001137883,"about_ca_system_score_gemma":0.003688657,"threshold_uncertainty_score":0.08064443},"labels":[],"label_agreement":null},{"id":"W2171600077","doi":"10.1093/biostatistics/kxp050","title":"Confidence intervals that match Fisher's exact or Blaker's exact tests","year":2009,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Statistical Methods and Models","field":"Mathematics","cited_by":151,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Exact test; Exact statistics; Confidence interval; Mathematics; Statistics; Credible interval; Coverage probability; Robust confidence intervals; CDF-based nonparametric confidence interval; Tolerance interval; Binomial proportion confidence interval; Confidence region; Confidence distribution; Poisson distribution","score_opus":0.1830855602638954,"score_gpt":0.4425573308257128,"score_spread":0.25947177056181736,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2171600077","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022800403,0.005139828,0.9465501,0.0015474979,0.0012459268,0.0005842182,0.0023897202,0.0033815936,0.016360858],"genre_scores_gemma":[0.38199735,0.0018874217,0.6047442,0.0015466299,0.0010493464,0.0023341146,0.0025947907,0.0015867855,0.002259391],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9264167,0.041142303,0.006196842,0.007879194,0.016667485,0.0016974831],"domain_scores_gemma":[0.50282747,0.4122987,0.02503884,0.03349835,0.024091175,0.002245457],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.059123695,0.0014054151,0.0027092013,0.008001457,0.000985578,0.004641282,0.0050123744,0.003607675,0.01831946],"category_scores_gemma":[0.57701635,0.0009095849,0.002450467,0.009796565,0.002951852,0.006796514,0.004713609,0.005482443,0.004237712],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023785348,0.0003609896,0.022805251,0.0033428518,0.0016032504,0.0011571073,0.0024962977,0.037148647,0.0020576732,0.30120257,0.054741867,0.570705],"study_design_scores_gemma":[0.0007223879,0.000987157,0.020461388,0.0024082996,0.0007959661,0.0027757094,0.0011645864,0.11628677,0.0071156486,0.7428464,0.1039145,0.00052117364],"about_ca_topic_score_codex":0.0017082668,"about_ca_topic_score_gemma":0.00092331745,"teacher_disagreement_score":0.059123695,"about_ca_system_score_codex":0.0012545801,"about_ca_system_score_gemma":0.0018615988,"threshold_uncertainty_score":0.3126797},"labels":[],"label_agreement":null},{"id":"W2198637531","doi":"10.1093/biostatistics/kxv028","title":"Double robust and efficient estimation of a prognostic model for events in the presence of dependent censoring","year":2015,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Institutes of Health; National Institute of Allergy and Infectious Diseases; ACT Government","keywords":"Censoring (clinical trials); Inverse probability weighting; Inverse probability; Estimator; Computer science; Inference; Statistics; Econometrics; Weighting; Missing data; Estimation; Causal inference; Parametric model; Parametric statistics; Mathematics; Artificial intelligence; Posterior probability","score_opus":0.2544352714323289,"score_gpt":0.4149531699672734,"score_spread":0.16051789853494447,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2198637531","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0042324886,0.00023146327,0.9947408,0.00023218113,0.000013841214,0.000027083506,0.0000833841,0.00015454523,0.00028424594],"genre_scores_gemma":[0.4029343,0.0010808114,0.59096164,0.00034226375,0.00017693856,0.00035097878,0.0009567115,0.00017386394,0.0030225143],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99412006,0.0039793137,0.00023719588,0.0007317845,0.00070493453,0.00022669723],"domain_scores_gemma":[0.9762241,0.017539417,0.002032281,0.0028910951,0.0010250392,0.00028813086],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.015466734,0.000983738,0.0018218032,0.0011728907,0.00038883393,0.0013600943,0.0026342233,0.0017379574,0.0022904626],"category_scores_gemma":[0.049556904,0.00071036565,0.0014526214,0.0011854396,0.0014036008,0.0026123556,0.0025971518,0.002497262,0.000634373],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042453085,0.00015193301,0.0070351134,0.00032890815,0.00037607708,0.00036893203,0.00025583338,0.6175855,0.0022506877,0.19349022,0.003720292,0.174012],"study_design_scores_gemma":[0.00003849406,0.00006248168,0.00076170184,0.000027726259,0.00004370631,0.000118396056,0.0000164976,0.90647054,0.0007465077,0.09024184,0.0014517626,0.000020389407],"about_ca_topic_score_codex":0.0028246671,"about_ca_topic_score_gemma":0.0022972757,"teacher_disagreement_score":0.015466734,"about_ca_system_score_codex":0.0009965887,"about_ca_system_score_gemma":0.0019947581,"threshold_uncertainty_score":0.081796885},"labels":[],"label_agreement":null},{"id":"W2254245873","doi":"10.1093/biostatistics/kxv015","title":"Methods to assess an exercise intervention trial based on 3-level functional data","year":2015,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Cancer Institute; Ministerio de Ciencia e Innovación; National Science Foundation","keywords":"Random effects model; Functional data analysis; Inference; Principal component analysis; Model selection; Functional principal component analysis; Mathematics; Computer science; Statistics; Artificial intelligence; Medicine; Meta-analysis","score_opus":0.7722899850188043,"score_gpt":0.5685272321347117,"score_spread":0.20376275288409262,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2254245873","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0134001365,0.0010115146,0.9623265,0.0009176686,0.0003599387,0.017278211,0.0023115918,0.0007938023,0.0016005913],"genre_scores_gemma":[0.14768946,0.00070002687,0.7195107,0.0009926258,0.0002122813,0.12597272,0.002103796,0.00016379364,0.002654586],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9159099,0.07209783,0.0032102526,0.0044174814,0.0036126452,0.00075199234],"domain_scores_gemma":[0.90106755,0.078126095,0.007899128,0.008294715,0.0034415033,0.0011710381],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.09112223,0.002309028,0.0037357465,0.0027513562,0.00067173317,0.0016584788,0.004171613,0.0027980397,0.019888585],"category_scores_gemma":[0.19199964,0.0012771836,0.005058966,0.0022329309,0.0015658455,0.0023306878,0.0027201432,0.0034428663,0.0017408619],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.03042119,0.0033829222,0.036744256,0.007419101,0.013447603,0.0005914639,0.0007824874,0.1022096,0.002163447,0.14917588,0.014544886,0.6391172],"study_design_scores_gemma":[0.014597389,0.009653051,0.019405223,0.0015045448,0.0054851775,0.00045167215,0.00016371718,0.6768074,0.002545423,0.24213423,0.026901638,0.00035045174],"about_ca_topic_score_codex":0.0015387293,"about_ca_topic_score_gemma":0.0020880876,"teacher_disagreement_score":0.09112223,"about_ca_system_score_codex":0.0014947221,"about_ca_system_score_gemma":0.0041701254,"threshold_uncertainty_score":0.48190618},"labels":[],"label_agreement":null},{"id":"W2259606755","doi":"10.1093/biostatistics/kxv054","title":"Augmented composite likelihood for copula modeling in family studies under biased sampling","year":2016,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; National Psoriasis Foundation","keywords":"Copula (linguistics); Statistics; Quasi-maximum likelihood; Marginal likelihood; Econometrics; Heritability; Maximum likelihood; Sampling bias; Sampling scheme; Marginal model; Importance sampling; Computer science; Mathematics; Sample size determination; Likelihood function; Regression analysis; Biology; Estimator; Genetics; Monte Carlo method","score_opus":0.4362143197409914,"score_gpt":0.46424727938488547,"score_spread":0.028032959643894073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2259606755","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040431716,0.00016606066,0.99520516,0.00012739745,0.000012072147,0.000029266703,0.0000631094,0.000084238956,0.00026958392],"genre_scores_gemma":[0.19509192,0.0010401057,0.7989578,0.0002204653,0.00014118326,0.0010227204,0.0005099202,0.00024676608,0.0027690905],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98600453,0.012204615,0.0002491423,0.0006914626,0.0006787008,0.00017148776],"domain_scores_gemma":[0.88677645,0.10303715,0.0034052061,0.0047279177,0.001489477,0.00056387496],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.027683867,0.001073312,0.0018252139,0.0015783109,0.0007555445,0.001759947,0.002797505,0.0015164309,0.0034899423],"category_scores_gemma":[0.10374729,0.00086736714,0.0015103179,0.0023614538,0.0024904665,0.002235515,0.002575542,0.0029182045,0.0006665584],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015774365,0.00006927809,0.010391823,0.00028721846,0.00038950998,0.00076761306,0.0008539985,0.3678134,0.00082861184,0.55490917,0.002425444,0.061106198],"study_design_scores_gemma":[0.000019376264,0.00003611749,0.0008168896,0.00003961343,0.000033323642,0.0001248014,0.00004909885,0.79427624,0.00017071656,0.20282106,0.0015886143,0.000024237639],"about_ca_topic_score_codex":0.005251951,"about_ca_topic_score_gemma":0.0043707695,"teacher_disagreement_score":0.027683867,"about_ca_system_score_codex":0.0012743748,"about_ca_system_score_gemma":0.0022850751,"threshold_uncertainty_score":0.14640802},"labels":[],"label_agreement":null},{"id":"W2271845266","doi":"10.1093/biostatistics/kxv042","title":"Sieve estimation in a Markov illness-death process under dual censoring","year":2015,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Censoring (clinical trials); Estimator; Statistics; Inference; Sieve (category theory); Imputation (statistics); Markov chain; Mathematics; Computer science; Parametric statistics; Econometrics; Missing data; Artificial intelligence","score_opus":0.15012141993189768,"score_gpt":0.4187033755424467,"score_spread":0.268581955610549,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2271845266","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0155035965,0.00018161548,0.98365754,0.00013186045,0.00001648057,0.000029979376,0.000050780378,0.000106024316,0.00032217283],"genre_scores_gemma":[0.55978155,0.0016194971,0.42947167,0.00037714007,0.00013572542,0.00073556305,0.0010142933,0.00014313332,0.0067214593],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972052,0.0016866163,0.00014531625,0.00041312052,0.00037608505,0.00017366189],"domain_scores_gemma":[0.9644333,0.03070671,0.0015643699,0.0017412582,0.00094773644,0.0006066354],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01365204,0.00079468946,0.003312495,0.0022402797,0.0005933673,0.0017355252,0.002560709,0.0025651702,0.0030263541],"category_scores_gemma":[0.049216244,0.0011103315,0.0025506017,0.0017897332,0.0024081455,0.003022238,0.0031854394,0.0036660174,0.00053164747],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038575756,0.00014489297,0.0067307157,0.00027088475,0.0003246283,0.00044694642,0.00045881327,0.60691404,0.0015907455,0.32274616,0.0011981062,0.05878834],"study_design_scores_gemma":[0.00002319257,0.00003013534,0.00045182192,0.000034005407,0.000022409828,0.00005503958,0.000015994336,0.92748094,0.0003270438,0.07105737,0.00048135014,0.000020705731],"about_ca_topic_score_codex":0.0017884942,"about_ca_topic_score_gemma":0.0015090541,"teacher_disagreement_score":0.01365204,"about_ca_system_score_codex":0.00082475005,"about_ca_system_score_gemma":0.0012591449,"threshold_uncertainty_score":0.07219976},"labels":[],"label_agreement":null},{"id":"W2320372797","doi":"10.1093/biostatistics/kxt057","title":"Monitoring risk-adjusted medical outcomes allowing for changes over time","year":2013,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Statistical Process Monitoring","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Chart; Computer science; Variety (cybernetics); Failure rate; Measure (data warehouse); Data mining; Statistics; Artificial intelligence; Mathematics","score_opus":0.10301405804493143,"score_gpt":0.4256607621804991,"score_spread":0.3226467041355677,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2320372797","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06272738,0.0002335689,0.9344865,0.0006457522,0.0000510421,0.000086954555,0.00045814144,0.0007473613,0.0005631625],"genre_scores_gemma":[0.7496777,0.00023210765,0.2480126,0.0001816286,0.00010488794,0.00018148696,0.0007354002,0.000097122494,0.0007769707],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9914596,0.004596354,0.00055890006,0.0018711231,0.0011474929,0.0003664234],"domain_scores_gemma":[0.9400175,0.041440815,0.01012367,0.005296445,0.0025592672,0.00056233257],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018681614,0.0012640083,0.0014120586,0.0023333237,0.00039527248,0.0017724067,0.0020371897,0.0017103318,0.0009460246],"category_scores_gemma":[0.089728504,0.00050089485,0.0014411109,0.0024291456,0.00076737395,0.0019156474,0.0013488393,0.0019439833,0.00023371205],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005282857,0.0002536256,0.11999975,0.00018615092,0.00062961975,0.00040738357,0.0004195344,0.6429795,0.0037482916,0.018091477,0.001353619,0.21140271],"study_design_scores_gemma":[0.000032876367,0.00019323586,0.013059752,0.000024416155,0.00008126541,0.00014574887,0.000028489,0.9655296,0.0018683607,0.01822211,0.0007597599,0.0000544816],"about_ca_topic_score_codex":0.006767687,"about_ca_topic_score_gemma":0.0040523573,"teacher_disagreement_score":0.018681614,"about_ca_system_score_codex":0.0008513323,"about_ca_system_score_gemma":0.001095079,"threshold_uncertainty_score":0.09879899},"labels":[],"label_agreement":null},{"id":"W2465707019","doi":"10.1093/biostatistics/kxw022","title":"Prediction of cancer drug sensitivity using high-dimensional omic features","year":2016,"lang":"en","type":"article","venue":"Biostatistics","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"National Institute of General Medical Sciences; National Cancer Institute","keywords":"Feature selection; Sensitivity (control systems); Curse of dimensionality; Computational biology; Omics; Computer science; Set (abstract data type); Personalized medicine; Machine learning; Bioinformatics; Artificial intelligence; Biology","score_opus":0.017586134820108467,"score_gpt":0.27013563471640223,"score_spread":0.25254949989629377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2465707019","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.31990948,0.0016204797,0.67406094,0.0008926189,0.00006150137,0.0002075149,0.0016617902,0.0008359007,0.00074984774],"genre_scores_gemma":[0.9152557,0.0003613525,0.08252173,0.00011799979,0.0000550997,0.00016330874,0.0012082779,0.000026458563,0.00028999374],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983475,0.00087796606,0.00013309083,0.00022867351,0.0003026273,0.00011016604],"domain_scores_gemma":[0.99446565,0.0040241973,0.0006268804,0.00035890818,0.00040701442,0.000117438445],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033463198,0.0008835437,0.0016483832,0.0028725062,0.00041857557,0.0012039263,0.0006338182,0.0008026379,0.00050444895],"category_scores_gemma":[0.008072737,0.0002708872,0.0018731448,0.0020064216,0.00042713442,0.00058491935,0.00083069387,0.00097283634,0.00016221385],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065496546,0.0005744437,0.10108736,0.00044911494,0.0013557033,0.00043641857,0.00013599507,0.65435493,0.013474703,0.0029970922,0.0024517418,0.22202748],"study_design_scores_gemma":[0.000031005402,0.00009398075,0.010172506,0.000015636042,0.000076993754,0.0000860532,0.000021692125,0.9828608,0.0017271371,0.0044832868,0.0004077099,0.000023088944],"about_ca_topic_score_codex":0.002246388,"about_ca_topic_score_gemma":0.0021183528,"teacher_disagreement_score":0.0033463198,"about_ca_system_score_codex":0.00059563183,"about_ca_system_score_gemma":0.00096822536,"threshold_uncertainty_score":0.017697275},"labels":[],"label_agreement":null},{"id":"W2470470131","doi":"10.1093/biostatistics/kxw026","title":"The positive effects of population-based preferential sampling in environmental epidemiology","year":2016,"lang":"en","type":"article","venue":"Biostatistics","topic":"Air Quality and Health Impacts","field":"Environmental Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Army Research Office; Defense Advanced Research Projects Agency; National Institute of Environmental Health Sciences; Natural Sciences and Engineering Research Council of Canada; U.S. Environmental Protection Agency; National Science Foundation","keywords":"Sampling (signal processing); Statistics; Population; Context (archaeology); Population density; Econometrics; Estimation; Environmental epidemiology; Sampling bias; Sampling design; Sample (material); Computer science; Environmental science; Sample size determination; Environmental health; Mathematics; Geography; Medicine","score_opus":0.0380394959602653,"score_gpt":0.3242569791752058,"score_spread":0.2862174832149405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2470470131","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06738273,0.0029819375,0.9201612,0.003401351,0.00015171093,0.0003164238,0.00037322752,0.0002606378,0.0049707135],"genre_scores_gemma":[0.76637006,0.0030564524,0.22563776,0.0023069351,0.00038721552,0.0004275096,0.0002857295,0.00005810125,0.0014702119],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9434179,0.043068953,0.0014293785,0.0063566114,0.004834043,0.0008930528],"domain_scores_gemma":[0.6475111,0.30127266,0.01858676,0.024406262,0.007066903,0.0011562352],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.050850853,0.0012902681,0.0019175091,0.0018381211,0.0013804184,0.0020746517,0.0031211432,0.003383473,0.0019133568],"category_scores_gemma":[0.19263422,0.001293522,0.0020202508,0.0035081543,0.0076224585,0.0066554314,0.004470393,0.0034117624,0.00029874637],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00083801436,0.0003751025,0.21959274,0.0012576102,0.0014888523,0.0012530335,0.0022693335,0.3839241,0.004372499,0.23450574,0.0035000446,0.14662296],"study_design_scores_gemma":[0.00022578755,0.0012520484,0.053755086,0.00037711946,0.00062432827,0.0035518778,0.0006522603,0.5583638,0.00576015,0.36643183,0.008744883,0.0002607866],"about_ca_topic_score_codex":0.010202937,"about_ca_topic_score_gemma":0.007939658,"teacher_disagreement_score":0.050850853,"about_ca_system_score_codex":0.002170275,"about_ca_system_score_gemma":0.002001219,"threshold_uncertainty_score":0.26892823},"labels":[],"label_agreement":null},{"id":"W2788646752","doi":"10.1093/biostatistics/kxy083","title":"Fast nonconvex deconvolution of calcium imaging data","year":2018,"lang":"en","type":"preprint","venue":"Biostatistics","topic":"Neural dynamics and brain function","field":"Neuroscience","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute on Drug Abuse; Canadian Institutes of Health Research; Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; National Science Foundation","keywords":"Deconvolution; Python (programming language); Computer science; Algorithm; Set (abstract data type); Artificial intelligence; Programming language","score_opus":0.10155622646636496,"score_gpt":0.3378078324847482,"score_spread":0.23625160601838324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2788646752","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01246921,0.00013522014,0.9845867,0.00043068486,0.000033628952,0.000037197984,0.00036623713,0.0009557831,0.0009853493],"genre_scores_gemma":[0.21155645,0.00031393467,0.77830076,0.00039867638,0.00009423646,0.00026535834,0.002610943,0.0007349902,0.00572471],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989961,0.00027354836,0.00006660563,0.00022813703,0.00033531766,0.00010034032],"domain_scores_gemma":[0.9963726,0.0022653847,0.00021095216,0.0005139278,0.00052657555,0.0001105995],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028264162,0.0012437606,0.0009987658,0.00069802016,0.0005728126,0.0010782887,0.001500292,0.0016354239,0.0029669139],"category_scores_gemma":[0.009777291,0.0005963308,0.0011650876,0.00095216953,0.0011214184,0.0014240451,0.0019519653,0.0023318855,0.0012147526],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019401751,0.000059219587,0.0007896714,0.00016265901,0.0000514467,0.00017454193,0.000087941975,0.88820595,0.008301976,0.018781962,0.006698921,0.0764917],"study_design_scores_gemma":[0.0000057987954,0.000005720922,0.0001165377,0.0000032171258,0.0000012630987,0.000014390039,0.000005474953,0.9935447,0.0012914805,0.0044225766,0.00058386166,0.00000494971],"about_ca_topic_score_codex":0.0115888985,"about_ca_topic_score_gemma":0.015081315,"teacher_disagreement_score":0.0115888985,"about_ca_system_score_codex":0.0012839534,"about_ca_system_score_gemma":0.0029042375,"threshold_uncertainty_score":0.023042917},"labels":[],"label_agreement":null},{"id":"W2795107686","doi":"10.1093/biostatistics/kxx036","title":"Lagged kernel machine regression for identifying time windows of susceptibility to exposures of complex mixtures","year":2017,"lang":"en","type":"article","venue":"Biostatistics","topic":"Air Quality Monitoring and Forecasting","field":"Environmental Science","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"FAS Division of Science, Harvard University; Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Environmental Health Sciences; National Science Foundation; National Institutes of Health; Harvard University","keywords":"Regression; Kernel (algebra); Kernel regression; Statistics; Computer science; Regression analysis; Econometrics; Machine learning; Artificial intelligence; Mathematics","score_opus":0.0819267791111343,"score_gpt":0.3532885025019369,"score_spread":0.27136172339080256,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2795107686","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07420837,0.001269883,0.921133,0.0004299564,0.00012121881,0.000057472964,0.00076377334,0.0012708171,0.0007454394],"genre_scores_gemma":[0.78177786,0.000892662,0.21086903,0.00024745738,0.00017835388,0.00018639222,0.0022322626,0.00025372874,0.0033622477],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99880886,0.00053441914,0.000072846175,0.00034930886,0.00014328494,0.00009134387],"domain_scores_gemma":[0.99608445,0.0024413797,0.0005611196,0.0003617388,0.00045959218,0.00009170584],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003035172,0.00093083916,0.00092843926,0.0008554025,0.00032161525,0.0006711816,0.0012654556,0.00093851157,0.0018053236],"category_scores_gemma":[0.010247717,0.00043332842,0.0012236492,0.00093721633,0.0004425731,0.00092334254,0.00087993365,0.0017299372,0.0005623595],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00049163186,0.00019018086,0.023364898,0.000306407,0.0004177872,0.0002646528,0.00012944605,0.8076261,0.0058061765,0.009615996,0.0033445905,0.14844224],"study_design_scores_gemma":[0.000007980185,0.00003502305,0.0020841018,0.000009189563,0.00001949704,0.000023433584,0.000011351382,0.993417,0.0005744342,0.0031296057,0.00067526766,0.000013108369],"about_ca_topic_score_codex":0.007048677,"about_ca_topic_score_gemma":0.004744648,"teacher_disagreement_score":0.007048677,"about_ca_system_score_codex":0.0005871153,"about_ca_system_score_gemma":0.00117288,"threshold_uncertainty_score":0.01605171},"labels":[],"label_agreement":null},{"id":"W2795350022","doi":"10.1093/biostatistics/kxx047","title":"A joint model for mixed and truncated longitudinal data and survival data, with application to HIV vaccine studies","year":2017,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health; Bill and Melinda Gates Foundation","keywords":"Inference; Computer science; Likelihood function; Missing data; Generalized estimating equation; Statistics; Longitudinal data; Statistical inference; Algorithm; Maximum likelihood; Mathematics; Data mining; Estimation theory; Artificial intelligence","score_opus":0.33641280286451325,"score_gpt":0.45578092942858783,"score_spread":0.11936812656407458,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2795350022","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008289065,0.0005707783,0.9888891,0.00090650015,0.000093313334,0.000105070896,0.00043756873,0.000122002864,0.00058651366],"genre_scores_gemma":[0.37782845,0.0028768282,0.600414,0.0012205781,0.00081544026,0.002276288,0.0022933185,0.00021822707,0.012056905],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9876881,0.0081767505,0.00059325283,0.0019656867,0.0010906773,0.00048556045],"domain_scores_gemma":[0.9547701,0.036747675,0.003453828,0.0026743151,0.0016508307,0.00070333347],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.028623102,0.001501858,0.0029205463,0.002167987,0.0009484848,0.0029414727,0.0047728983,0.0030997647,0.004268964],"category_scores_gemma":[0.051128894,0.001369812,0.0029095418,0.0030389603,0.003072412,0.0049517695,0.0034336073,0.004366711,0.00072305027],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002512719,0.00011917563,0.009677092,0.00046479612,0.00048894354,0.00069111737,0.000665604,0.34894013,0.0008832869,0.5895461,0.0031895686,0.045082938],"study_design_scores_gemma":[0.000048625483,0.00010769687,0.0013990791,0.00007256454,0.000100609075,0.00016201555,0.00007989045,0.7725739,0.00020388371,0.22237259,0.0028286956,0.000050385508],"about_ca_topic_score_codex":0.006344156,"about_ca_topic_score_gemma":0.0056162113,"teacher_disagreement_score":0.028623102,"about_ca_system_score_codex":0.0020230187,"about_ca_system_score_gemma":0.002708655,"threshold_uncertainty_score":0.15137523},"labels":[],"label_agreement":null},{"id":"W2890289581","doi":"10.1093/biostatistics/kxy041","title":"Pointless spatial modeling","year":2018,"lang":"en","type":"article","venue":"Biostatistics","topic":"Spatial and Panel Data Analysis","field":"Economics, Econometrics and Finance","cited_by":36,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Cancer Institute; Medical Research Council; National Institutes of Health; Applied Molecular Biosciences Unit; Kurdistan University Of Medical Sciences; Mekelle University; Đại học Quốc gia Hà Nội; University of Peradeniya; Addis Ababa University; University of Gondar; University of Tabriz; Universidade Federal de Sergipe; Universitatea de Medicină şi Farmacie \"Carol Davila\" Bucureşti; Universidade do Porto; Bahir Dar University; Alexandria University; Bill and Melinda Gates Foundation; Karolinska Institutet; Tabriz University of Medical Sciences; Shahroud University of Medical Sciences; Babol University of Medical Sciences; Tehran University of Medical Sciences and Health Services; Mazandaran University of Medical Sciences; Universität Bielefeld; Aksum University; Public Health Foundation of India; Kaiser Permanente; Australian Catholic University; Mansoura University; Hamadan University of Medical Sciences; University of Oxford; Universidad Autónoma de Sinaloa; Maragheh University of Medical Sciences; Universidad Nacional Autónoma de México; Indian Institute of Technology Delhi; Istituto di Ricerche Farmacologiche Mario Negri - IRCCS; University of Southampton; University of Washington; A.T. Still University; Simon Fraser University; University of Ottawa; U.S. Department of Veterans Affairs","keywords":"Computer science; Laplace's method; Bayesian inference; Smoothing; Random field; Markov random field; Bayesian probability; Variable-order Bayesian network; Algorithm; Theoretical computer science; Applied mathematics; Mathematics; Artificial intelligence; Statistics","score_opus":0.059526377990475644,"score_gpt":0.23771309163695656,"score_spread":0.17818671364648092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2890289581","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008605618,0.0006533894,0.97801197,0.0011711745,0.00011523047,0.000065037886,0.0015396692,0.00034584594,0.0094921375],"genre_scores_gemma":[0.66509956,0.0034178465,0.27483946,0.00093032507,0.0007464665,0.0006910056,0.00353101,0.0003958458,0.05034847],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9981281,0.00069897505,0.0001038469,0.0004939869,0.00041904696,0.00015595958],"domain_scores_gemma":[0.99579793,0.0023130535,0.000628754,0.0006262036,0.00045890996,0.0001750617],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028626132,0.0010192348,0.0017242851,0.002832578,0.00072553207,0.00319428,0.0050893435,0.0030936056,0.011487262],"category_scores_gemma":[0.010971731,0.000894147,0.0021350547,0.0041454304,0.0023180014,0.004505812,0.002984673,0.0021797628,0.0023875057],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001788011,0.000018633908,0.001392029,0.00005890841,0.00004438912,0.00011566388,0.00012610576,0.23897228,0.00016243092,0.74626905,0.0028501602,0.009972544],"study_design_scores_gemma":[0.000012120224,0.000015546699,0.00027652032,0.000023423043,0.000018489647,0.00008381733,0.00003209652,0.7509012,0.000064112,0.24264503,0.0059125526,0.000015124553],"about_ca_topic_score_codex":0.013480738,"about_ca_topic_score_gemma":0.008046471,"teacher_disagreement_score":0.013480738,"about_ca_system_score_codex":0.0019853634,"about_ca_system_score_gemma":0.0011561888,"threshold_uncertainty_score":0.038428783},"labels":[],"label_agreement":null},{"id":"W2910244130","doi":"10.1093/biostatistics/kxz001","title":"Are clusterings of multiple data views independent?","year":2019,"lang":"en","type":"preprint","venue":"Biostatistics","topic":"Metabolomics and Mass Spectrometry Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of General Medical Sciences; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; National Science Foundation","keywords":"Cluster analysis; Spurious relationship; Data type; Computer science; Data set; Set (abstract data type); Exploit; Data mining; Data science; Information retrieval; Psychology; Machine learning; Artificial intelligence","score_opus":0.07597877536538193,"score_gpt":0.3191912477251362,"score_spread":0.24321247235975424,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2910244130","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4130882,0.0017192191,0.5590969,0.0069201495,0.00076697563,0.0017392809,0.00401095,0.0009148878,0.011743517],"genre_scores_gemma":[0.92880386,0.0003436274,0.061704762,0.0013433492,0.00051141676,0.0008217942,0.0046660574,0.00032785686,0.0014774094],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8185756,0.08788162,0.009999094,0.057618923,0.021597002,0.0043277354],"domain_scores_gemma":[0.41193524,0.3928586,0.053599034,0.11381236,0.022712575,0.0050823125],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.13413605,0.0020075506,0.0043288544,0.0040248553,0.0037025851,0.008917959,0.0068513695,0.0052013653,0.0051509095],"category_scores_gemma":[0.39051867,0.002293793,0.0082723955,0.0046393503,0.014341627,0.011223353,0.007104823,0.008407049,0.0019549476],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006179141,0.0010375659,0.6942118,0.0014317462,0.015986882,0.0017490685,0.011755071,0.022292651,0.006070089,0.09624461,0.011150472,0.13189097],"study_design_scores_gemma":[0.0011140124,0.0011361687,0.46618295,0.0005458597,0.002419825,0.0025523114,0.004582805,0.21410365,0.00676813,0.28929427,0.010607997,0.00069203996],"about_ca_topic_score_codex":0.004820782,"about_ca_topic_score_gemma":0.0028511987,"teacher_disagreement_score":0.13413605,"about_ca_system_score_codex":0.0029706624,"about_ca_system_score_gemma":0.0026806043,"threshold_uncertainty_score":0.7093877},"labels":[],"label_agreement":null},{"id":"W2941651377","doi":"10.1093/biostatistics/kxz011","title":"Parallel repulsive logic regression with biological adjacency","year":2019,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Cancer Institute; National Institutes of Health","keywords":"Adjacency list; Adjacency matrix; Computer science; SNP; Regression; Function (biology); Similarity (geometry); Mathematics; Single-nucleotide polymorphism; Algorithm; Theoretical computer science; Artificial intelligence; Biology; Statistics; Genetics; Graph","score_opus":0.009814861069642562,"score_gpt":0.23850521024042726,"score_spread":0.2286903491707847,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2941651377","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014698511,0.00008221404,0.9834425,0.00015543461,0.000022076685,0.000041129122,0.00004808839,0.000513372,0.0009967359],"genre_scores_gemma":[0.45841408,0.00020283906,0.53476316,0.00048788972,0.00009902075,0.00033395033,0.00045878586,0.00030428747,0.0049360106],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9988537,0.00046551603,0.000056992252,0.00019535444,0.00033229895,0.00009606851],"domain_scores_gemma":[0.9966112,0.002488856,0.0003370596,0.00015477247,0.00032624367,0.00008187195],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021507055,0.0010882134,0.0010594083,0.001144421,0.00059595495,0.0009972844,0.0019064859,0.000914965,0.00289677],"category_scores_gemma":[0.0069038975,0.0004949145,0.00091421074,0.0011463069,0.00087232847,0.001382991,0.0016443032,0.0012456101,0.00059804745],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016750465,0.00008611759,0.0020759841,0.00009279342,0.00005961323,0.0001523779,0.000050311428,0.8518056,0.0038925984,0.02464443,0.0012027783,0.115769856],"study_design_scores_gemma":[0.000007902048,0.0000143888665,0.000043902048,0.0000016875853,0.0000035636585,0.00001251682,0.000002893528,0.9952095,0.0003998728,0.004142083,0.0001583152,0.0000034250263],"about_ca_topic_score_codex":0.004005008,"about_ca_topic_score_gemma":0.004518516,"teacher_disagreement_score":0.004005008,"about_ca_system_score_codex":0.00080946356,"about_ca_system_score_gemma":0.0014344335,"threshold_uncertainty_score":0.011374116},"labels":[],"label_agreement":null},{"id":"W2963817914","doi":"10.1093/biostatistics/kxy049","title":"Bayesian model averaging for the X-chromosome inactivation dilemma in genetic association study","year":2018,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"National Human Genome Research Institute; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Bayesian probability; Association (psychology); Dilemma; Genetics; Chromosome; Genetic association; Genetic model; Computer science; Biology; Computational biology; Statistics; Econometrics; Mathematics; Artificial intelligence; Psychology; Genotype; Gene; Single-nucleotide polymorphism","score_opus":0.018205598656295813,"score_gpt":0.2923468390028399,"score_spread":0.2741412403465441,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2963817914","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036294546,0.00031809797,0.9952915,0.00029817008,0.000024230552,0.000025883845,0.000027194495,0.000066156055,0.00031932804],"genre_scores_gemma":[0.26795006,0.0013901255,0.72677857,0.00056236784,0.0004088774,0.0006001824,0.0002674427,0.0001424401,0.0018998958],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9855186,0.011899094,0.00038795624,0.0009338788,0.0010702603,0.00019025135],"domain_scores_gemma":[0.93237495,0.06175495,0.0018147107,0.002568997,0.0010304481,0.00045597486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.033470463,0.0010814588,0.002279736,0.0014117131,0.0010588141,0.0019485765,0.0028840099,0.0020006495,0.0017248407],"category_scores_gemma":[0.1010117,0.0009382479,0.001483252,0.0020484545,0.0032725346,0.0028396337,0.0026055374,0.0031350548,0.0003505213],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013095989,0.000053464202,0.0030931672,0.00022650315,0.00035400657,0.0003505514,0.0004179792,0.46404007,0.0010036649,0.47696394,0.0019711475,0.051394492],"study_design_scores_gemma":[0.000030261224,0.000036907666,0.0004277677,0.000025606798,0.000040869705,0.00008302713,0.000022403594,0.74742,0.00019854287,0.25063872,0.001043869,0.000031970303],"about_ca_topic_score_codex":0.0062671457,"about_ca_topic_score_gemma":0.00560389,"teacher_disagreement_score":0.033470463,"about_ca_system_score_codex":0.0012458859,"about_ca_system_score_gemma":0.0021771064,"threshold_uncertainty_score":0.17701083},"labels":[],"label_agreement":null},{"id":"W2986102733","doi":"10.1093/biostatistics/kxz047","title":"Independence conditions and the analysis of life history studies with intermittent observation","year":2019,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Independence (probability theory); Econometrics; Statistics; Computer science; Psychology; Mathematics","score_opus":0.15193314052114745,"score_gpt":0.370486172072266,"score_spread":0.21855303155111852,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2986102733","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027999356,0.0016469787,0.96430415,0.0015829446,0.00010801113,0.00018586754,0.0007738811,0.00020449115,0.0031943673],"genre_scores_gemma":[0.69387335,0.0050440985,0.27805218,0.0012211505,0.0010626896,0.0040178145,0.003682133,0.00032924575,0.01271734],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9773907,0.016238673,0.0009991126,0.0028030602,0.0018401152,0.0007283365],"domain_scores_gemma":[0.5655918,0.39667478,0.019338975,0.012381329,0.0037776427,0.0022354627],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.06315823,0.001324126,0.002427191,0.0042966288,0.0014966234,0.002888383,0.0038853735,0.0030303132,0.01012319],"category_scores_gemma":[0.22714873,0.001641536,0.0037969674,0.00315226,0.0073352256,0.0062025087,0.0049622664,0.0047934675,0.00088983204],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018494566,0.00014561058,0.023171071,0.00040037298,0.0006495876,0.0011063158,0.0010633746,0.059049,0.0006593551,0.8785695,0.0017477096,0.033253238],"study_design_scores_gemma":[0.00008533245,0.00016994598,0.0071240067,0.00018758385,0.00016484474,0.00033623996,0.00019851027,0.20787919,0.0005461027,0.7788023,0.004434991,0.00007084612],"about_ca_topic_score_codex":0.0050067315,"about_ca_topic_score_gemma":0.002284874,"teacher_disagreement_score":0.06315823,"about_ca_system_score_codex":0.0018751321,"about_ca_system_score_gemma":0.0030242559,"threshold_uncertainty_score":0.33401662},"labels":[],"label_agreement":null},{"id":"W2990086743","doi":"10.1093/biostatistics/kxz048","title":"The illness-death model for family studies","year":2019,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Covariate; Family history; Disease; Psoriatic arthritis; Sampling (signal processing); Construct (python library); Medicine; Sampling bias; Statistics; Missing data; Demography; Computer science; Sample size determination; Mathematics; Internal medicine","score_opus":0.24790220017718542,"score_gpt":0.4473781306493679,"score_spread":0.19947593047218246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990086743","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013431661,0.0023849842,0.9727447,0.003860052,0.0002726226,0.00044215214,0.0027157436,0.00023056778,0.0039175292],"genre_scores_gemma":[0.4386903,0.007991942,0.49789894,0.0026718108,0.0014315514,0.00786473,0.008673574,0.00029999219,0.03447722],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98818386,0.009189651,0.00038379565,0.0011537453,0.000753251,0.00033576027],"domain_scores_gemma":[0.93310237,0.05621311,0.003654767,0.0046430267,0.0015100493,0.0008767749],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.036098078,0.0011801072,0.0020917202,0.0025705185,0.0012286069,0.002130715,0.0048654685,0.0029236989,0.016984839],"category_scores_gemma":[0.08061423,0.0008413522,0.0019608971,0.0037956797,0.0024363524,0.0034860673,0.0026999076,0.0043794997,0.0018174328],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013232243,0.00007204421,0.012087478,0.00033393854,0.00036666743,0.0007480401,0.0007088308,0.063344084,0.00017575205,0.88267064,0.009249889,0.030110272],"study_design_scores_gemma":[0.00017562644,0.00009624499,0.0027461061,0.00016514839,0.0001289487,0.00044377186,0.0001830875,0.26253414,0.000092011534,0.7170345,0.016351542,0.000048843805],"about_ca_topic_score_codex":0.01105996,"about_ca_topic_score_gemma":0.010140803,"teacher_disagreement_score":0.036098078,"about_ca_system_score_codex":0.002277477,"about_ca_system_score_gemma":0.0029413789,"threshold_uncertainty_score":0.19090718},"labels":[],"label_agreement":null},{"id":"W2993029184","doi":"10.1093/biostatistics/kxz053","title":"Pair-based likelihood approximations for stochastic epidemic models","year":2019,"lang":"en","type":"article","venue":"Biostatistics","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Engineering and Physical Sciences Research Council","keywords":"Markov chain Monte Carlo; Computer science; Approximate Bayesian computation; Markov chain; Mathematical optimization; Epidemic model; Algorithm; Machine learning; Artificial intelligence; Mathematics; Bayesian probability; Inference","score_opus":0.2613787351654007,"score_gpt":0.41705189516470365,"score_spread":0.15567315999930292,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2993029184","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003801144,0.0003965995,0.99289274,0.0003398305,0.000033925135,0.000037299113,0.0001295029,0.00017094954,0.00219813],"genre_scores_gemma":[0.38804075,0.0021085273,0.58794963,0.0005317533,0.0004096855,0.0009533179,0.0018076344,0.0005832064,0.017615411],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966425,0.002159225,0.00011683641,0.00028113666,0.00061710394,0.00018328906],"domain_scores_gemma":[0.9795685,0.017039692,0.00084081123,0.0011618542,0.0010075686,0.00038165424],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008903861,0.0012520005,0.0020482945,0.00215898,0.0007940813,0.0024792382,0.004198157,0.0026756197,0.006213052],"category_scores_gemma":[0.044995245,0.0011835898,0.001823242,0.0023843742,0.0021161474,0.0041590533,0.0033561192,0.004204904,0.0018799647],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059634214,0.0000448789,0.001536254,0.00013130823,0.000067162524,0.00022168084,0.0002005793,0.618656,0.00024136523,0.35810402,0.0029720073,0.017765196],"study_design_scores_gemma":[0.000008055199,0.0000074221416,0.00008488135,0.0000147771,0.000004704905,0.000031725103,0.000016066064,0.9115651,0.00004764078,0.087139934,0.0010720567,0.0000076028064],"about_ca_topic_score_codex":0.0075201057,"about_ca_topic_score_gemma":0.004933039,"teacher_disagreement_score":0.008903861,"about_ca_system_score_codex":0.002180993,"about_ca_system_score_gemma":0.0016661381,"threshold_uncertainty_score":0.047088683},"labels":[],"label_agreement":null},{"id":"W3010419042","doi":"10.1093/biostatistics/kxaa009","title":"Geographically dependent individual-level models for infectious diseases transmission","year":2020,"lang":"en","type":"article","venue":"Biostatistics","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto; University of Calgary","funders":"Canadian Statistical Sciences Institute; Natural Sciences and Engineering Research Council of Canada","keywords":"Transmission (telecommunications); Computer science; Econometrics; Statistics; Mathematics","score_opus":0.31380845844225735,"score_gpt":0.39699155322202456,"score_spread":0.08318309477976721,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3010419042","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06877294,0.003984656,0.8920281,0.0053997478,0.0005246295,0.00038236103,0.007327807,0.0008945303,0.020685248],"genre_scores_gemma":[0.82120997,0.00678524,0.09596392,0.0013127708,0.0005921257,0.0012110985,0.0054025142,0.0002977585,0.067224525],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9976299,0.0011729882,0.0001126342,0.0005784061,0.00028459964,0.000221432],"domain_scores_gemma":[0.9958746,0.0022851394,0.00082666427,0.0003184999,0.00046723548,0.00022788228],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0032849326,0.0015808743,0.0017403201,0.0019274317,0.00086176896,0.0025225072,0.0062296377,0.003465511,0.011944092],"category_scores_gemma":[0.009265562,0.0010824166,0.002053123,0.0026692953,0.0017822244,0.003007727,0.0024158452,0.0036936407,0.0030172493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007749358,0.000110526336,0.008415814,0.00016554762,0.00025105526,0.0004074081,0.00051677146,0.74050057,0.00039755282,0.23023924,0.0055985292,0.013319541],"study_design_scores_gemma":[0.000045005334,0.000055140692,0.0020506214,0.000043323787,0.00008439917,0.0001501933,0.0001239957,0.9108061,0.000056327863,0.080964684,0.0055718813,0.00004827401],"about_ca_topic_score_codex":0.039261397,"about_ca_topic_score_gemma":0.02460142,"teacher_disagreement_score":0.039261397,"about_ca_system_score_codex":0.0026483599,"about_ca_system_score_gemma":0.0015335096,"threshold_uncertainty_score":0.07806575},"labels":[],"label_agreement":null},{"id":"W3011802598","doi":"10.1093/biostatistics/kxaa011","title":"Assessing the accuracy of predictive models with interval-censored data","year":2020,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Canadian Institutes of Health Research; University of Waterloo; National Natural Science Foundation of China","keywords":"Estimator; Inverse probability; Statistics; Computer science; Event (particle physics); Imputation (statistics); Context (archaeology); Receiver operating characteristic; Inverse probability weighting; Coverage probability; Confidence interval; Mathematics; Bayesian probability; Missing data; Posterior probability","score_opus":0.337170611191294,"score_gpt":0.44963140921643074,"score_spread":0.11246079802513675,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3011802598","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032880917,0.00092142785,0.96466345,0.00038416527,0.00004472738,0.00007964178,0.00015323973,0.0003205358,0.0005519072],"genre_scores_gemma":[0.69577116,0.0013649806,0.30013534,0.0002592114,0.00025168547,0.0003517064,0.0009974918,0.000117271644,0.0007511712],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98033285,0.013900456,0.00093995105,0.0016348682,0.0028387285,0.00035316063],"domain_scores_gemma":[0.7400014,0.23259518,0.011601168,0.010011214,0.0050080637,0.00078300084],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.053741287,0.001521284,0.0018153974,0.003652509,0.0006235654,0.0023545488,0.0030254032,0.002046536,0.0008496143],"category_scores_gemma":[0.2207416,0.00070630026,0.001557252,0.0024135294,0.0020672537,0.00244494,0.0024235547,0.0027702926,0.000346549],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003078877,0.00018737296,0.05682815,0.00035471318,0.0011780395,0.00015832242,0.0003709896,0.77452224,0.0009624561,0.025912045,0.0011224723,0.13809536],"study_design_scores_gemma":[0.000016329066,0.000114327486,0.003244872,0.00006479728,0.000050241266,0.000060981634,0.00003823605,0.97913057,0.0004925109,0.016379125,0.0003753539,0.000032713604],"about_ca_topic_score_codex":0.00475741,"about_ca_topic_score_gemma":0.0029675544,"teacher_disagreement_score":0.053741287,"about_ca_system_score_codex":0.0008958179,"about_ca_system_score_gemma":0.0020249863,"threshold_uncertainty_score":0.2842145},"labels":[],"label_agreement":null},{"id":"W3015827596","doi":"10.1093/biostatistics/kxaa012","title":"Two-part joint model for a longitudinal semicontinuous marker and a terminal event with application to metastatic colorectal cancer data","year":2020,"lang":"en","type":"article","venue":"Biostatistics","topic":"Colorectal Cancer Treatments and Studies","field":"Medicine","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; University of Toronto; Mount Sinai Hospital","funders":"National Cancer Institute; Canadian Institutes of Health Research","keywords":"Biomarker; Event (particle physics); Colorectal cancer; Proportional hazards model; Outcome (game theory); Oncology; Computer science; Statistics; Cancer; Medicine; Internal medicine; Mathematics; Biology; Physics","score_opus":0.0662028208738954,"score_gpt":0.3436616069784594,"score_spread":0.277458786104564,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3015827596","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21582268,0.0033163545,0.7699697,0.003199935,0.00032371315,0.0006519226,0.003587597,0.0006627697,0.0024653692],"genre_scores_gemma":[0.88441044,0.0016796287,0.08992614,0.00076683593,0.00041083523,0.0014442249,0.004719217,0.00017954208,0.01646311],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9898773,0.006722731,0.00039152039,0.0015486698,0.00068897964,0.0007708101],"domain_scores_gemma":[0.948377,0.041769713,0.003542717,0.0026699582,0.0026238468,0.0010166854],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.029961996,0.0022652952,0.0039681704,0.0016638396,0.0010489393,0.0029491477,0.004162924,0.0033194104,0.005518431],"category_scores_gemma":[0.033443183,0.0011799437,0.0051723765,0.0023163985,0.002518236,0.002436072,0.0025404114,0.004930629,0.0010635047],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0024034928,0.0004814978,0.046642363,0.00046859065,0.0014853001,0.00252247,0.0010553626,0.7986165,0.0016169793,0.09963636,0.003831851,0.041239277],"study_design_scores_gemma":[0.00016798703,0.000404074,0.0064199124,0.000053497093,0.000314758,0.00024539974,0.00012717494,0.9619539,0.00027687295,0.027940307,0.002009814,0.00008634714],"about_ca_topic_score_codex":0.018581094,"about_ca_topic_score_gemma":0.01103013,"teacher_disagreement_score":0.029961996,"about_ca_system_score_codex":0.0021283622,"about_ca_system_score_gemma":0.0027272746,"threshold_uncertainty_score":0.15845603},"labels":[],"label_agreement":null},{"id":"W3021772758","doi":"10.1093/biostatistics/kxy075","title":"Sequence kernel association test for survival outcomes in the presence of a non-susceptible fraction","year":2018,"lang":"en","type":"article","venue":"Biostatistics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lunenfeld-Tanenbaum Research Institute; University of Toronto; Sinai Health System; Université Laval","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Sequence (biology); Association (psychology); Fraction (chemistry); Statistics; Test (biology); Kernel (algebra); Computer science; Mathematics; Psychology; Biology; Genetics; Combinatorics; Chemistry","score_opus":0.02798905792391057,"score_gpt":0.3321394866421573,"score_spread":0.3041504287182467,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3021772758","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30002874,0.0003203044,0.69624203,0.0006497593,0.000095189855,0.000076870965,0.00040619465,0.00048454248,0.001696387],"genre_scores_gemma":[0.9437218,0.00008146972,0.054930925,0.00010100516,0.000057465444,0.00006833483,0.00033933952,0.000033743327,0.0006659763],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9921051,0.004413024,0.00045943903,0.0013815803,0.0013506504,0.00029016804],"domain_scores_gemma":[0.93218607,0.05762099,0.0035931636,0.003925036,0.0020552417,0.00061956566],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011016013,0.00038545698,0.0012664329,0.0021795854,0.0004845527,0.0013371604,0.0020183146,0.0010973607,0.002618392],"category_scores_gemma":[0.058962572,0.00025535433,0.001229335,0.0014725561,0.001670712,0.0016754509,0.0016654053,0.0014791702,0.00044250226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022519429,0.00067279884,0.39053813,0.00042552172,0.001993076,0.0016892795,0.0005809809,0.12833577,0.009439766,0.13125052,0.0022704797,0.3305518],"study_design_scores_gemma":[0.00028556414,0.00083763566,0.06473507,0.000058340724,0.0006801681,0.0013062707,0.00021558988,0.8133125,0.0045083943,0.1115483,0.0023886634,0.00012350803],"about_ca_topic_score_codex":0.00116555,"about_ca_topic_score_gemma":0.0011823047,"teacher_disagreement_score":0.011016013,"about_ca_system_score_codex":0.0004919664,"about_ca_system_score_gemma":0.0015384613,"threshold_uncertainty_score":0.05825895},"labels":[],"label_agreement":null},{"id":"W3026963084","doi":"10.1093/biostatistics/kxaa020","title":"Causal inference for recurrent event data using pseudo-observations","year":2020,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; HEC Montréal; Jewish General Hospital","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research; Institut de Valorisation des Données","keywords":"Estimator; Inverse probability weighting; Statistics; Inverse probability; Mathematics; Weighting; Regression; Event (particle physics); Regression analysis; Computer science; Econometrics; Bayesian probability; Medicine","score_opus":0.5969398403880808,"score_gpt":0.49798059002811046,"score_spread":0.09895925035997033,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3026963084","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029206034,0.00033201097,0.9692616,0.00013916918,0.000058309044,0.00013323089,0.00033164307,0.00020209062,0.0003360278],"genre_scores_gemma":[0.6874553,0.00060030224,0.30749834,0.000267153,0.00015645713,0.0009535627,0.0019363427,0.000104398496,0.0010280613],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9744124,0.019262541,0.0010984156,0.0028674917,0.0020291042,0.00033018005],"domain_scores_gemma":[0.7524452,0.20883383,0.014605972,0.019311626,0.004203004,0.0006003196],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.050335396,0.00090107234,0.001638335,0.0025842378,0.000769216,0.001985697,0.003230982,0.0017556412,0.0034866503],"category_scores_gemma":[0.23650107,0.00063764903,0.0021193437,0.0024931594,0.0023744917,0.0033363837,0.0023393955,0.0024629391,0.00036753644],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001381979,0.00042599544,0.11509507,0.0012416763,0.001999536,0.0012246657,0.0015479202,0.21548593,0.0021889787,0.34432283,0.0028525114,0.31223288],"study_design_scores_gemma":[0.00014220526,0.00047588412,0.020139713,0.00016937726,0.00030549758,0.0004689264,0.00023118693,0.7082088,0.0022155575,0.263731,0.0038090886,0.00010272016],"about_ca_topic_score_codex":0.001985456,"about_ca_topic_score_gemma":0.0013676943,"teacher_disagreement_score":0.050335396,"about_ca_system_score_codex":0.00074119045,"about_ca_system_score_gemma":0.0010611841,"threshold_uncertainty_score":0.26620215},"labels":[],"label_agreement":null},{"id":"W3027831021","doi":"10.1093/biostatistics/kxab016","title":"Bayesian integrative analysis and prediction with application to atherosclerosis cardiovascular disease","year":2021,"lang":"en","type":"preprint","venue":"Biostatistics","topic":"Genetic Associations and Epidemiology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"National Center for Advancing Translational Sciences; National Institutes of Health","keywords":"Disease; Atherosclerotic cardiovascular disease; Genome-wide association study; Computational biology; Genetic variants; Risk factor; Bioinformatics; Medicine; Gene; Biology; Genetics; Internal medicine; Single-nucleotide polymorphism; Genotype","score_opus":0.00791382655362262,"score_gpt":0.23941766280060273,"score_spread":0.2315038362469801,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3027831021","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.021365615,0.0010130563,0.97412574,0.0011238185,0.000073080635,0.00005984205,0.0001665081,0.0003385845,0.001733613],"genre_scores_gemma":[0.5561183,0.0023785164,0.43146968,0.00053474965,0.00045509997,0.00050046534,0.0008512804,0.00032605304,0.0073657962],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974099,0.001793362,0.000077244564,0.00027080777,0.00033621333,0.000112383525],"domain_scores_gemma":[0.9869485,0.011129519,0.00049388374,0.00042344973,0.000761795,0.00024288987],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008929,0.0010360779,0.0018525948,0.001237296,0.0008461516,0.0013069025,0.001500637,0.001314476,0.0025389872],"category_scores_gemma":[0.02698897,0.0008345008,0.0014679625,0.0016580856,0.0017568584,0.0015080436,0.002344804,0.0024088968,0.00043221677],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008201425,0.00005585882,0.0030070262,0.00009328677,0.00015860579,0.00010751411,0.00017565707,0.82353646,0.00037343684,0.12612608,0.00187765,0.04440647],"study_design_scores_gemma":[0.000013437828,0.000011832757,0.00028742553,0.000009876197,0.000011922207,0.000011031941,0.00000841156,0.9393737,0.00006309428,0.059674833,0.00052483834,0.000009479838],"about_ca_topic_score_codex":0.024284152,"about_ca_topic_score_gemma":0.015428514,"teacher_disagreement_score":0.024284152,"about_ca_system_score_codex":0.001483074,"about_ca_system_score_gemma":0.0024916988,"threshold_uncertainty_score":0.048285604},"labels":[],"label_agreement":null},{"id":"W3080340456","doi":"10.1093/biostatistics/kxaa033","title":"Adaptive treatment strategies for chronic conditions: shared-parameter G-estimation with an application to rheumatoid arthritis","year":2020,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Fonds de Recherche du Québec - Santé","keywords":"Estimation; Computer science; Robustness (evolution); Rheumatoid arthritis; Mathematical optimization; Machine learning; Estimation theory; Point estimation; Artificial intelligence; Econometrics; Data mining; Medicine; Mathematics; Algorithm; Statistics","score_opus":0.09362554207553095,"score_gpt":0.3750745637280131,"score_spread":0.2814490216524822,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3080340456","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015529976,0.0005205833,0.9822305,0.00072026375,0.000055185315,0.00018520348,0.000095437914,0.00014672431,0.0005161963],"genre_scores_gemma":[0.47859648,0.0010244764,0.5147833,0.0008736388,0.0002844472,0.0011844747,0.00042760445,0.0001389044,0.0026866582],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9844381,0.01305129,0.00034864547,0.001232828,0.0006153125,0.00031379674],"domain_scores_gemma":[0.897812,0.09290408,0.0031806754,0.0042158486,0.0013582857,0.0005291251],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.030787885,0.0013149894,0.0028170792,0.0014215484,0.00078285555,0.0014294004,0.0035240296,0.0022721414,0.0039413837],"category_scores_gemma":[0.10173141,0.00089807715,0.0026264077,0.0017454296,0.0030702194,0.0025183212,0.00292344,0.004160983,0.00042693003],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008414155,0.00032959215,0.011387185,0.0003163804,0.0012357166,0.0005492116,0.0006238725,0.64900494,0.00076301757,0.16912581,0.0035521144,0.16227078],"study_design_scores_gemma":[0.00021660051,0.00018425507,0.0013021607,0.000057596295,0.00010530003,0.000070487884,0.000063069936,0.8101525,0.00024701524,0.18630724,0.0012531388,0.000040686777],"about_ca_topic_score_codex":0.008866653,"about_ca_topic_score_gemma":0.005449363,"teacher_disagreement_score":0.030787885,"about_ca_system_score_codex":0.001416618,"about_ca_system_score_gemma":0.0030605549,"threshold_uncertainty_score":0.16282386},"labels":[],"label_agreement":null},{"id":"W3186606527","doi":"10.1093/biostatistics/kxab028","title":"Bayesian adaptive model selection design for optimal biological dose finding in phase I/II clinical trials","year":2021,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Cancer Institute","keywords":"Bayesian probability; Selection (genetic algorithm); Computer science; Model selection; Maximum tolerated dose; Adaptive design; Econometrics; Clinical trial; Statistics; Artificial intelligence; Mathematics; Medicine; Internal medicine","score_opus":0.8908728524028277,"score_gpt":0.6664937251343948,"score_spread":0.2243791272684329,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3186606527","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009187951,0.00015724916,0.9891522,0.00020136753,0.000022961382,0.00045521953,0.000045104614,0.0001308353,0.00064709835],"genre_scores_gemma":[0.41281953,0.0002445916,0.5810785,0.00050166214,0.00006383407,0.0034661102,0.00017147718,0.00006337767,0.0015908511],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98324096,0.013827127,0.00044450833,0.0010651426,0.0010651923,0.00035704134],"domain_scores_gemma":[0.9872114,0.009234871,0.0015227325,0.00072563626,0.0008713147,0.0004340038],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.018425995,0.0012630612,0.0017785617,0.00085401477,0.0003691493,0.0009430157,0.0016751894,0.001652727,0.0028493365],"category_scores_gemma":[0.022545582,0.0008366868,0.0009997208,0.00077169726,0.0014607608,0.00094436307,0.0014482271,0.0022621793,0.0005412047],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030298957,0.0005025092,0.0025963606,0.00041998795,0.00026938558,0.00015880386,0.0002306151,0.76305836,0.009684965,0.06831036,0.0016991369,0.15003976],"study_design_scores_gemma":[0.000901147,0.001577917,0.00096034195,0.000047974572,0.00009634191,0.00006688918,0.000021610866,0.9612928,0.00293236,0.029632308,0.0024171101,0.000053245512],"about_ca_topic_score_codex":0.00063960976,"about_ca_topic_score_gemma":0.0006056923,"teacher_disagreement_score":0.018425995,"about_ca_system_score_codex":0.001128533,"about_ca_system_score_gemma":0.0022117125,"threshold_uncertainty_score":0.09744716},"labels":[],"label_agreement":null},{"id":"W3193894044","doi":"10.1093/biostatistics/kxab031","title":"Estimation for the bivariate quantile varying coefficient model with application to diffusion tensor imaging data analysis","year":2021,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Institutes of Health; National Cancer Institute; National Institute of Mental Health; Canadian Statistical Sciences Institute; Natural Sciences and Engineering Research Council of Canada; University of Alberta","keywords":"Interpretability; Bivariate analysis; Computer science; Quantile; Diffusion MRI; Quantile regression; Data set; Artificial intelligence; Mathematics; Econometrics; Machine learning","score_opus":0.085489773200644,"score_gpt":0.39188917805165296,"score_spread":0.30639940485100897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3193894044","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007793241,0.00014022161,0.9914239,0.00023961894,0.0000113315255,0.00002419043,0.00007365872,0.0001508634,0.00014305708],"genre_scores_gemma":[0.36065337,0.0012796421,0.63251114,0.00025515174,0.00011907396,0.00045729018,0.00086128246,0.00043916146,0.0034238694],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99775535,0.0015023106,0.000071193135,0.00031531372,0.00021674378,0.00013913747],"domain_scores_gemma":[0.9867025,0.010597913,0.00085720257,0.0007345704,0.000911575,0.00019620372],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01034898,0.00078507845,0.001203145,0.0011447273,0.0004220884,0.001081039,0.0020495593,0.0014665624,0.0021753055],"category_scores_gemma":[0.03232351,0.00072686805,0.0013374852,0.0016330532,0.0011450693,0.0013558348,0.0016610294,0.002609167,0.00050392485],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013855954,0.00006137762,0.007274163,0.0001509094,0.00016173317,0.00026537938,0.0002261233,0.80830634,0.0025522101,0.10341648,0.002445938,0.07500079],"study_design_scores_gemma":[0.0000088738025,0.0000140853035,0.0005363326,0.000010689975,0.000010515324,0.00003437998,0.000014269736,0.98062,0.00022248493,0.017887777,0.0006278786,0.0000127461435],"about_ca_topic_score_codex":0.010972941,"about_ca_topic_score_gemma":0.0071777413,"teacher_disagreement_score":0.010972941,"about_ca_system_score_codex":0.0009561882,"about_ca_system_score_gemma":0.001923201,"threshold_uncertainty_score":0.05473125},"labels":[],"label_agreement":null},{"id":"W3196024289","doi":"10.1093/biostatistics/kxab032","title":"Predicting the onset of breast cancer using mammogram imaging data with irregular boundary","year":2021,"lang":"en","type":"article","venue":"Biostatistics","topic":"AI in cancer detection","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"National Cancer Institute; Natural Sciences and Engineering Research Council of Canada; Breast Cancer Research Foundation","keywords":"Breast cancer; Computer science; Artificial intelligence; Mammography; Pattern recognition (psychology); Benchmark (surveying); Breast cancer screening; Boundary (topology); Machine learning; Medicine; Cancer; Mathematics","score_opus":0.02508658907445413,"score_gpt":0.2886221632118421,"score_spread":0.26353557413738793,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196024289","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71509016,0.00054518273,0.2813292,0.0003724289,0.000043023054,0.00012963508,0.00096237625,0.0004718773,0.0010560965],"genre_scores_gemma":[0.9470254,0.00020502853,0.05139912,0.000031895735,0.000032182244,0.00004600096,0.0008804768,0.00002199032,0.00035789405],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99960417,0.00014274055,0.00003435064,0.000085230284,0.000089659734,0.000043980035],"domain_scores_gemma":[0.9968262,0.0016343353,0.0007439144,0.00032118947,0.00035161682,0.00012283055],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009819954,0.00047792925,0.0005233268,0.0021166943,0.00021001462,0.00067492365,0.00039802937,0.0006567286,0.0005899553],"category_scores_gemma":[0.0071975146,0.00026085362,0.00055570976,0.0010750818,0.00036635285,0.00049678737,0.00053531857,0.00069418113,0.00024153612],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005812092,0.00028649397,0.61630905,0.00012962427,0.00018104604,0.00077728485,0.00015023477,0.19608508,0.016860712,0.0020869505,0.0016351523,0.16491719],"study_design_scores_gemma":[0.000014399245,0.00012758368,0.17235701,0.000030013838,0.00004339107,0.0005444549,0.00014342254,0.817877,0.0041968827,0.003899144,0.000731615,0.000034975474],"about_ca_topic_score_codex":0.005077488,"about_ca_topic_score_gemma":0.008787931,"teacher_disagreement_score":0.005077488,"about_ca_system_score_codex":0.00028632782,"about_ca_system_score_gemma":0.0004501793,"threshold_uncertainty_score":0.010095835},"labels":[],"label_agreement":null},{"id":"W4210676319","doi":"10.1093/biostatistics/kxac007","title":"Semiparametric Bayesian inference for optimal dynamic treatment regimes via dynamic marginal structural models","year":2022,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; ViiV Healthcare; Gilead Sciences","keywords":"Frequentist inference; Inference; Bayesian probability; Computer science; Bayesian inference; Posterior probability; Econometrics; Marginal structural model; Machine learning; Causal inference; Artificial intelligence; Mathematics","score_opus":0.05764133595221096,"score_gpt":0.379869733491707,"score_spread":0.32222839753949606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210676319","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012427948,0.0002505746,0.9851248,0.0007535215,0.000016852096,0.000060307673,0.00020744481,0.00012324103,0.0010352797],"genre_scores_gemma":[0.67279625,0.0011089175,0.32111952,0.00045491633,0.0001260527,0.0005786619,0.0007512974,0.00013094739,0.0029334654],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9925223,0.0054963953,0.00023139239,0.0008376231,0.00061248423,0.00029968913],"domain_scores_gemma":[0.9587752,0.0366934,0.0021441672,0.0013428354,0.000698419,0.00034587982],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014905237,0.00084144523,0.002241478,0.0017175137,0.000580644,0.0023340508,0.0024496384,0.0016189773,0.0043600756],"category_scores_gemma":[0.06466967,0.00104157,0.0017791494,0.00159329,0.0026356156,0.003288871,0.0023909516,0.003889177,0.00045365316],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014164577,0.00008848561,0.002454612,0.00015516444,0.00023123797,0.00009766761,0.0002338357,0.5825897,0.00033256918,0.3763045,0.0014006087,0.03596997],"study_design_scores_gemma":[0.000033170618,0.000022909857,0.00043049222,0.000034243378,0.000031119955,0.000019600642,0.00002633925,0.7305437,0.00013485302,0.26808092,0.0006257511,0.000016904545],"about_ca_topic_score_codex":0.0060276883,"about_ca_topic_score_gemma":0.0075105526,"teacher_disagreement_score":0.014905237,"about_ca_system_score_codex":0.002573087,"about_ca_system_score_gemma":0.002351336,"threshold_uncertainty_score":0.07882732},"labels":[],"label_agreement":null},{"id":"W4229024550","doi":"10.1093/biostatistics/kxac029","title":"A flexible approach for predictive biomarker discovery","year":2022,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Covariate; Computer science; False positive paradox; Estimator; Inference; Identification (biology); Nonparametric statistics; False positives and false negatives; Metric (unit); Biomarker discovery; Multiple comparisons problem; Predictive inference; Biomarker; Data mining; False discovery rate; Truncation (statistics); Machine learning; Artificial intelligence; Statistics; Frequentist inference; Mathematics; Bayesian probability; Bayesian inference","score_opus":0.5564742844991701,"score_gpt":0.5379670415846817,"score_spread":0.018507242914488464,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4229024550","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006524356,0.00009251655,0.998447,0.00022182755,0.000018355517,0.000037788315,0.000052546002,0.00016677297,0.000310812],"genre_scores_gemma":[0.11198158,0.0004405372,0.8831693,0.00064190826,0.00023400568,0.0007135077,0.00035950282,0.00023042194,0.002229232],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9830351,0.009513656,0.0007598345,0.0027582333,0.0034348252,0.0004983268],"domain_scores_gemma":[0.9576892,0.02950291,0.0025094734,0.007950409,0.00176337,0.00058456085],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.026413832,0.0014686798,0.001886032,0.0031382516,0.001079683,0.0030608445,0.0047898977,0.002143033,0.00443176],"category_scores_gemma":[0.083425276,0.0012674116,0.0026709768,0.0038964788,0.0037364406,0.0035826662,0.006220057,0.005926491,0.0015068414],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023196786,0.00011892843,0.003152674,0.0003065358,0.00035289224,0.00046614438,0.0003757586,0.1896771,0.0036258053,0.5502614,0.004672894,0.24675797],"study_design_scores_gemma":[0.00007151887,0.000086552864,0.0007407783,0.000072380826,0.00006442886,0.00018807974,0.000024565747,0.47899726,0.0017904273,0.510402,0.007497201,0.00006485787],"about_ca_topic_score_codex":0.0016772824,"about_ca_topic_score_gemma":0.0015514791,"teacher_disagreement_score":0.026413832,"about_ca_system_score_codex":0.0016188688,"about_ca_system_score_gemma":0.0029637953,"threshold_uncertainty_score":0.13969135},"labels":[],"label_agreement":null},{"id":"W4230814105","doi":"10.1093/biostatistics/5.3.361","title":"Analysis of longitudinal marginal structural models","year":2004,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Marginal structural model; Econometrics; Marginal model; Computer science; Statistics; Mathematics; Regression analysis; Causal inference","score_opus":0.08612267330736914,"score_gpt":0.3841181678001392,"score_spread":0.2979954944927701,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4230814105","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.045334183,0.0006548362,0.95142514,0.0011536482,0.000049330192,0.00005343147,0.00036015152,0.00031284685,0.0006564078],"genre_scores_gemma":[0.72601265,0.0019542095,0.2630853,0.00035980297,0.0002981441,0.00062299846,0.0014637051,0.00029651084,0.0059066913],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9920587,0.0062187356,0.00017242985,0.0008057264,0.00046016823,0.00028427574],"domain_scores_gemma":[0.9005547,0.09060265,0.002782887,0.0037295774,0.0014781078,0.0008521166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.021848205,0.00097478705,0.0016872999,0.0017628645,0.0008436494,0.0020682171,0.002789063,0.0014821372,0.008254713],"category_scores_gemma":[0.09654679,0.0011186848,0.0023480768,0.0016034426,0.0017296516,0.002988005,0.002298241,0.0028330819,0.00067324284],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000705726,0.00031794034,0.02429232,0.0003930962,0.0013288562,0.00037814328,0.00089347246,0.2029542,0.0010464966,0.6282952,0.0052208686,0.13417365],"study_design_scores_gemma":[0.00006444907,0.00007722664,0.0027273728,0.00004711738,0.00015235867,0.000089507696,0.00008864926,0.5903964,0.00028649892,0.40479216,0.00124773,0.00003052106],"about_ca_topic_score_codex":0.005591903,"about_ca_topic_score_gemma":0.0060310094,"teacher_disagreement_score":0.021848205,"about_ca_system_score_codex":0.0014578882,"about_ca_system_score_gemma":0.0035071094,"threshold_uncertainty_score":0.11554569},"labels":[],"label_agreement":null},{"id":"W4283823477","doi":"10.1093/biostatistics/kxac023","title":"Constrained groupwise additive index models","year":2022,"lang":"en","type":"article","venue":"Biostatistics","topic":"Health, Environment, Cognitive Aging","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ouranos; University of Ottawa; Health Canada; Institut National de Santé Publique du Québec; Institut National de la Recherche Scientifique","funders":"Mitacs; Institut National de Santé Publique du Québec","keywords":"Computer science; Context (archaeology); Index (typography); Variance (accounting); Predictive power; Variable (mathematics); Inference; Data mining; Sensitivity (control systems); Econometrics; Statistics; Mathematics; Artificial intelligence","score_opus":0.017002022804621625,"score_gpt":0.23908750135535406,"score_spread":0.22208547855073243,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283823477","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0084994715,0.0004526785,0.988059,0.00031839614,0.00010634121,0.00014023633,0.00061917974,0.00024061254,0.0015640056],"genre_scores_gemma":[0.36685216,0.0022433621,0.6079249,0.00089659845,0.000554797,0.0020867873,0.0033327101,0.00023208148,0.015876666],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9933478,0.004129756,0.00024096109,0.001047517,0.00087686576,0.00035703604],"domain_scores_gemma":[0.981938,0.013591472,0.0015392572,0.00135388,0.0012918259,0.00028553532],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010271239,0.0029183545,0.0035427746,0.0026524605,0.000815239,0.0026211105,0.0056645684,0.0025065027,0.0068006483],"category_scores_gemma":[0.027528165,0.0010471471,0.002642721,0.004572004,0.0022603353,0.003089075,0.002941804,0.0036935515,0.0017565407],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002880085,0.00019725485,0.0053010243,0.00034853216,0.0005279353,0.00025961743,0.00033516544,0.5482672,0.0007775065,0.33106697,0.0046247398,0.10800604],"study_design_scores_gemma":[0.000035693534,0.00011404276,0.0005908248,0.00004114123,0.00008560858,0.00006110507,0.000029110826,0.82308495,0.00029872867,0.17286979,0.0027455187,0.000043500615],"about_ca_topic_score_codex":0.0051456857,"about_ca_topic_score_gemma":0.0053269677,"teacher_disagreement_score":0.010271239,"about_ca_system_score_codex":0.0013154993,"about_ca_system_score_gemma":0.0018521204,"threshold_uncertainty_score":0.054320157},"labels":[],"label_agreement":null},{"id":"W4311239350","doi":"10.1093/biostatistics/kxac047","title":"Inference after latent variable estimation for single-cell RNA sequencing data","year":2022,"lang":"en","type":"article","venue":"Biostatistics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of General Medical Sciences; National Institute on Drug Abuse; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health; Simons Foundation","keywords":"Inference; Latent variable; Count data; Computer science; Type I and type II errors; Latent variable model; Context (archaeology); Poisson distribution; Statistical inference; Data set; Data mining; Algorithm; Latent class model; Statistics; Mathematics; Machine learning; Artificial intelligence; Biology","score_opus":0.04935330434777099,"score_gpt":0.2670004868434502,"score_spread":0.2176471824956792,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311239350","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006109238,0.00006467892,0.9933344,0.00012161582,0.00001994265,0.0000358979,0.00005229324,0.00013691797,0.00012490075],"genre_scores_gemma":[0.3262846,0.00034925717,0.6691582,0.0004464448,0.0001911515,0.0007181915,0.0011851285,0.00025470674,0.001412403],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9849823,0.010082923,0.00061164616,0.0023800929,0.0014927986,0.00045020806],"domain_scores_gemma":[0.87595946,0.110799715,0.0034254105,0.007280181,0.0019854452,0.00054976414],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.035113137,0.0010283049,0.0019552342,0.0021414787,0.0013520161,0.0025566127,0.0032682756,0.0019551578,0.002478691],"category_scores_gemma":[0.1341296,0.0011080005,0.0020520175,0.0025031047,0.0037048322,0.0041213334,0.0028013547,0.004855841,0.0005870126],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007768848,0.00031794564,0.030659696,0.00051182136,0.0009358615,0.0007082641,0.0010408049,0.3323841,0.0070269573,0.4465559,0.0033727789,0.17570907],"study_design_scores_gemma":[0.00006541268,0.000053802894,0.0013636696,0.000031445503,0.000038823644,0.00007676299,0.000063918604,0.75527006,0.0020004064,0.24032287,0.0006854245,0.000027305792],"about_ca_topic_score_codex":0.003762474,"about_ca_topic_score_gemma":0.0034619558,"teacher_disagreement_score":0.035113137,"about_ca_system_score_codex":0.0013129045,"about_ca_system_score_gemma":0.0021734547,"threshold_uncertainty_score":0.18569827},"labels":[],"label_agreement":null},{"id":"W4321611071","doi":"10.1093/biostatistics/kxad002","title":"Assessing the causal effects of a stochastic intervention in time series data: are heat alerts effective in preventing deaths and hospitalizations?","year":2023,"lang":"en","type":"review","venue":"Biostatistics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute on Minority Health and Health Disparities; National Institute of Environmental Health Sciences; Wellcome Trust; Alfred P. Sloan Foundation; National Institute on Aging; National Institutes of Health; Harvard University","keywords":"Causal inference; Estimator; Context (archaeology); Computer science; Time series; Inference; Econometrics; Series (stratigraphy); Nonparametric statistics; Statistics; Machine learning; Artificial intelligence; Mathematics","score_opus":0.13287255036280732,"score_gpt":0.4694060812927,"score_spread":0.3365335309298927,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321611071","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13412493,0.004516642,0.8468956,0.0073833656,0.00071975746,0.0015768786,0.001570105,0.00031099422,0.0029017872],"genre_scores_gemma":[0.8805243,0.0020090668,0.11072529,0.0012465214,0.00056479144,0.0027100965,0.0007382261,0.00003745905,0.0014442288],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.95095843,0.038478322,0.0024884653,0.0050675375,0.002326513,0.0006807379],"domain_scores_gemma":[0.78782177,0.17799495,0.017797988,0.013620108,0.0019741368,0.00079112116],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.060981378,0.001123747,0.002661876,0.002381963,0.0010422375,0.0023896578,0.0032654232,0.0031940797,0.0050011124],"category_scores_gemma":[0.16534628,0.00079226843,0.0052803396,0.0031248024,0.0031829495,0.0034882743,0.0030748814,0.0038998818,0.0002972919],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023330702,0.0014171199,0.23612884,0.0051471484,0.018558795,0.00070044753,0.0012452091,0.17222014,0.0015764459,0.37641174,0.0035225265,0.18073861],"study_design_scores_gemma":[0.0011977488,0.0037564058,0.06383262,0.0013982865,0.012317438,0.00038191662,0.000875482,0.3761005,0.0034502882,0.5231094,0.013341558,0.00023841829],"about_ca_topic_score_codex":0.004548769,"about_ca_topic_score_gemma":0.0024278034,"teacher_disagreement_score":0.9390186,"about_ca_system_score_codex":0.0015256229,"about_ca_system_score_gemma":0.0031889465,"threshold_uncertainty_score":0.32250422},"labels":[],"label_agreement":null},{"id":"W4323350268","doi":"10.1093/biostatistics/kxad003","title":"A Bayesian approach to estimating COVID-19 incidence and infection fatality rates","year":2023,"lang":"en","type":"article","venue":"Biostatistics","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of British Columbia; Centre for Global Health Research; St. Michael's Hospital; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bayesian probability; Coronavirus disease 2019 (COVID-19); Statistics; Medicine; Proxy (statistics); Case fatality rate; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Multivariate statistics; Pandemic; Econometrics; Computer science; Mathematics; Disease; Epidemiology; Infectious disease (medical specialty); Internal medicine","score_opus":0.2973987214825874,"score_gpt":0.47669000304276166,"score_spread":0.17929128156017426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323350268","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012201428,0.00026916843,0.98541486,0.00044990305,0.000020390566,0.000087911045,0.00037010558,0.00017341254,0.0010127999],"genre_scores_gemma":[0.2967724,0.00095359713,0.69561744,0.00029194824,0.00020063767,0.00063609984,0.0016302528,0.000101137615,0.0037965022],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99413234,0.0037791955,0.00033655748,0.0007829028,0.00076549075,0.00020354816],"domain_scores_gemma":[0.98559177,0.010863589,0.0012067575,0.0011231495,0.0009966242,0.00021810344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01301249,0.00070043444,0.0012954935,0.0034141368,0.0008857125,0.0019123362,0.0025226357,0.0015171372,0.0027603973],"category_scores_gemma":[0.041158855,0.0011586506,0.001445764,0.002517883,0.0012615786,0.0022678098,0.0015368989,0.0026362787,0.0005386139],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019065452,0.00014738317,0.02773502,0.00020289564,0.00047770047,0.0002197157,0.000648374,0.5656533,0.0011451071,0.23387775,0.004093181,0.1656089],"study_design_scores_gemma":[0.000027464454,0.000049560538,0.0056328224,0.00007348431,0.00005277901,0.00014424803,0.00008100213,0.85709506,0.00027831836,0.13389874,0.0026177613,0.00004867148],"about_ca_topic_score_codex":0.034236882,"about_ca_topic_score_gemma":0.043780785,"teacher_disagreement_score":0.034236882,"about_ca_system_score_codex":0.0023996604,"about_ca_system_score_gemma":0.0024867142,"threshold_uncertainty_score":0.06881744},"labels":[],"label_agreement":null},{"id":"W4386404572","doi":"10.1093/biostatistics/kxad022","title":"Variable selection in high dimensions for discrete-outcome individualized treatment rules: Reducing severity of depression symptoms","year":2023,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; McGill University","funders":"National Institute of Mental Health; National Institutes of Health","keywords":"Depression (economics); Outcome (game theory); Feature selection; Selection (genetic algorithm); Variable (mathematics); Econometrics; Medicine; Statistics; Computer science; Internal medicine; Artificial intelligence; Mathematics; Mathematical economics; Economics","score_opus":0.10470571346149896,"score_gpt":0.41863948485896896,"score_spread":0.31393377139747003,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386404572","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09116226,0.00020512768,0.90636027,0.0009639319,0.000036029414,0.00018509751,0.000090562826,0.0001962207,0.0008005991],"genre_scores_gemma":[0.7306621,0.0001858306,0.26746905,0.00028068712,0.00007005323,0.0003464123,0.00027035305,0.000039191695,0.0006764324],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9907314,0.007529833,0.00027343386,0.000621517,0.00062938896,0.00021444188],"domain_scores_gemma":[0.95521986,0.03895224,0.0024517195,0.0020242308,0.00096371374,0.0003882461],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0138613405,0.00049123884,0.001478385,0.0008148821,0.0004944696,0.0011292143,0.0015669462,0.0010126092,0.0016710662],"category_scores_gemma":[0.054812565,0.0004562411,0.00086597237,0.00079442904,0.0009795952,0.001117104,0.0015184423,0.0019996585,0.00018143332],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063596753,0.0007030197,0.02876829,0.00024678817,0.0004864789,0.00027002455,0.0005115008,0.64780843,0.001593824,0.06280566,0.0029211768,0.25324884],"study_design_scores_gemma":[0.00009971081,0.00011903632,0.001994263,0.000022778031,0.000040292838,0.000025974061,0.000041546074,0.97129434,0.0005290021,0.025414834,0.0004035285,0.000014769865],"about_ca_topic_score_codex":0.003112072,"about_ca_topic_score_gemma":0.003679612,"teacher_disagreement_score":0.0138613405,"about_ca_system_score_codex":0.0007458996,"about_ca_system_score_gemma":0.0016039173,"threshold_uncertainty_score":0.07330668},"labels":[],"label_agreement":null},{"id":"W4399476155","doi":"10.1093/biostatistics/kxae016","title":"Fast matrix completion in epigenetic methylation studies with informative covariates","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"Epigenetics and DNA Methylation","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; HEC Montréal","funders":"Institut de Valorisation des Données","keywords":"Covariate; Missing data; Imputation (statistics); DNA methylation; Epigenetics; Leverage (statistics); Computer science; Statistics; Computational biology; Mathematics; Data mining; Biology; Genetics; Gene; Gene expression","score_opus":0.020573476340981032,"score_gpt":0.3349433991800419,"score_spread":0.31436992283906084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399476155","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0063312287,0.0006180551,0.99121195,0.000446051,0.000049231847,0.000088290275,0.0003182716,0.0006022611,0.0003345137],"genre_scores_gemma":[0.17316422,0.0014946804,0.81327605,0.00049098313,0.0002642246,0.0010891925,0.0036853566,0.00042700153,0.006108339],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9954933,0.002961621,0.0001883169,0.0006566553,0.00046129132,0.00023874626],"domain_scores_gemma":[0.98174816,0.0142012695,0.0009561105,0.0012543732,0.0012805855,0.0005595083],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009532434,0.0017924944,0.0023135757,0.0011991891,0.00089218834,0.0014060664,0.0027632432,0.002117469,0.0048039565],"category_scores_gemma":[0.02727506,0.0016641571,0.00202171,0.0023351791,0.0016917831,0.0022981698,0.0030451366,0.004707863,0.0017925074],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000682486,0.00022395754,0.004291656,0.00071554957,0.00029623156,0.0005219499,0.0004033882,0.787808,0.0025973064,0.0638331,0.007536769,0.13108958],"study_design_scores_gemma":[0.000050798088,0.00006386832,0.00041040077,0.000025258834,0.00001799382,0.000044062566,0.000026884145,0.96885407,0.000545461,0.027982157,0.0019551548,0.000023803044],"about_ca_topic_score_codex":0.018304426,"about_ca_topic_score_gemma":0.020687351,"teacher_disagreement_score":0.018304426,"about_ca_system_score_codex":0.0014507738,"about_ca_system_score_gemma":0.0042221644,"threshold_uncertainty_score":0.050412953},"labels":[],"label_agreement":null},{"id":"W4400446963","doi":"10.1093/biostatistics/kxae019","title":"A marginal structural model for normal tissue complication probability","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"Colorectal and Anal Carcinomas","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; Public Health Ontario; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; University of Toronto; Sanofi","keywords":"Radiation therapy; Context (archaeology); Estimator; Marginal structural model; Medicine; Computer science; Mathematics; Statistics; Radiology; Pathology; Causal inference","score_opus":0.04386924663779295,"score_gpt":0.32703391561408884,"score_spread":0.2831646689762959,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400446963","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07512182,0.0005046629,0.91767496,0.0018861167,0.000063653555,0.00019368477,0.001059276,0.00040083722,0.003095043],"genre_scores_gemma":[0.89166826,0.0008529382,0.096673824,0.0003300527,0.00017087183,0.0006480565,0.0010764088,0.00009991715,0.008479745],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9961582,0.0021234383,0.00014670068,0.00069469656,0.0004990528,0.00037789182],"domain_scores_gemma":[0.9714857,0.022426229,0.002509459,0.001727531,0.0013815397,0.0004696115],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009304973,0.0008361664,0.0014784392,0.0014947918,0.00055548234,0.0017861428,0.003677006,0.0019068288,0.008807153],"category_scores_gemma":[0.03590553,0.0006898419,0.001797717,0.0015198223,0.0026092401,0.0023989757,0.0016541941,0.0024922534,0.0009573902],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031832565,0.00016271281,0.018926783,0.00015310956,0.0002211953,0.00029755558,0.0005060671,0.51174873,0.0009802812,0.42660135,0.0028219835,0.03726193],"study_design_scores_gemma":[0.00009054653,0.00016708956,0.0045869034,0.00003901685,0.00011795233,0.00017517901,0.00008249649,0.8162844,0.00031807186,0.17623365,0.0018508327,0.000053896118],"about_ca_topic_score_codex":0.011788898,"about_ca_topic_score_gemma":0.008801353,"teacher_disagreement_score":0.011788898,"about_ca_system_score_codex":0.0021118592,"about_ca_system_score_gemma":0.0019103133,"threshold_uncertainty_score":0.049209952},"labels":[],"label_agreement":null},{"id":"W4403798551","doi":"10.1093/biostatistics/kxae039","title":"Selection processes, transportability, and failure time analysis in life history studies","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"Demographic Trends and Gender Preferences","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; University of Toronto","keywords":"Selection (genetic algorithm); Computer science; Econometrics; Artificial intelligence; Mathematics","score_opus":0.04926000744727137,"score_gpt":0.3216940800327122,"score_spread":0.27243407258544083,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403798551","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030842014,0.0010292208,0.9645948,0.0013171438,0.00011417979,0.00045975705,0.00030273237,0.0001316512,0.0012085224],"genre_scores_gemma":[0.77924585,0.0021327257,0.20548761,0.00075474626,0.0006096206,0.0040989025,0.0011368676,0.00023014095,0.006303571],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.90317625,0.07953584,0.0037676275,0.0073608025,0.0043745316,0.0017850539],"domain_scores_gemma":[0.39742228,0.54172474,0.027465945,0.025566852,0.006301093,0.0015189928],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.16776963,0.0024017,0.0032303696,0.006944696,0.0024977897,0.004285104,0.0053331438,0.0046361634,0.007171037],"category_scores_gemma":[0.38320377,0.0016306397,0.005984858,0.0057583433,0.011925502,0.008566398,0.006919074,0.006330346,0.0006680501],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042417034,0.00014687776,0.07066591,0.00059441006,0.0012475112,0.0013782819,0.003158284,0.1131033,0.00051773107,0.7695079,0.0012668425,0.037988715],"study_design_scores_gemma":[0.00013297002,0.00041380868,0.011233463,0.00025107324,0.0003638399,0.00045995234,0.0005348636,0.26807016,0.00051750586,0.7148311,0.0030430856,0.00014821194],"about_ca_topic_score_codex":0.00948638,"about_ca_topic_score_gemma":0.00300437,"teacher_disagreement_score":0.16776963,"about_ca_system_score_codex":0.0039253957,"about_ca_system_score_gemma":0.0035812194,"threshold_uncertainty_score":0.8872612},"labels":[],"label_agreement":null},{"id":"W4404260614","doi":"10.1093/biostatistics/kxae043","title":"Fast standard error estimation for joint models of longitudinal and time-to-event data based on stochastic EM algorithms","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"Distributed Sensor Networks and Detection Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Allergy and Infectious Diseases","keywords":"Computer science; Algorithm; Joint (building); Event (particle physics); Estimation; Event data; Data mining","score_opus":0.04732250925768137,"score_gpt":0.30083655796217923,"score_spread":0.25351404870449784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404260614","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014473598,0.000074531235,0.9981357,0.000052603464,0.000008403055,0.000017420072,0.00002651094,0.00013666986,0.000100882506],"genre_scores_gemma":[0.0946276,0.0004346536,0.90221024,0.00014280897,0.000070473565,0.00048176068,0.0006362141,0.00020548534,0.0011908009],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99410844,0.0033850065,0.00041718985,0.0009994892,0.0008998205,0.0001901126],"domain_scores_gemma":[0.95627546,0.038046144,0.0015969126,0.002124129,0.0017350977,0.00022227876],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.017665396,0.001544113,0.001996653,0.002284249,0.0006391588,0.0015536745,0.003404052,0.0015514392,0.0020489993],"category_scores_gemma":[0.067539014,0.0013040564,0.0017571177,0.0020455555,0.0016537689,0.0029744725,0.0029325876,0.003386177,0.00072848453],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015214119,0.00008890411,0.00423258,0.00024178655,0.0003164478,0.00011103491,0.00018268298,0.7273312,0.0016778985,0.10206692,0.0015109528,0.16208737],"study_design_scores_gemma":[0.000018613004,0.000024485522,0.00033115692,0.000023255408,0.000012851041,0.000039021023,0.000014849859,0.9579699,0.00082580344,0.040026315,0.00070003065,0.000013582702],"about_ca_topic_score_codex":0.0035455886,"about_ca_topic_score_gemma":0.0036564972,"teacher_disagreement_score":0.017665396,"about_ca_system_score_codex":0.0012958588,"about_ca_system_score_gemma":0.0031175031,"threshold_uncertainty_score":0.09342468},"labels":[],"label_agreement":null},{"id":"W4405942876","doi":"10.1093/biostatistics/kxae046","title":"Testing for a difference in means of a single feature after clustering","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of General Medical Sciences; National Institute on Drug Abuse; Natural Sciences and Engineering Research Council of Canada; National Institutes of Health","keywords":"Cluster analysis; Type I and type II errors; Feature (linguistics); Computer science; Hierarchical clustering; Data mining; Statistical hypothesis testing; Algorithm; Word error rate; Pattern recognition (psychology); Statistics; Artificial intelligence; Mathematics","score_opus":0.023888730904485327,"score_gpt":0.2522287114570061,"score_spread":0.22833998055252078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405942876","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25206965,0.0002594799,0.7430206,0.0006088022,0.0002530594,0.00030473075,0.00081215857,0.0008316781,0.0018398712],"genre_scores_gemma":[0.86039037,0.0000628309,0.13708973,0.00028582005,0.00009743224,0.0003883712,0.0011144129,0.00018169092,0.0003893287],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9724855,0.01133491,0.0018685914,0.008371822,0.004879198,0.0010599996],"domain_scores_gemma":[0.8490755,0.120600276,0.008278171,0.014735973,0.0058970093,0.001412945],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.030461183,0.0009477526,0.0022030596,0.0026958007,0.0015265965,0.0028037534,0.0025099872,0.003323819,0.0029910752],"category_scores_gemma":[0.12597717,0.00048871705,0.002015226,0.0023178845,0.0054743737,0.0029565066,0.0026166195,0.0030662315,0.00088620296],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0064666173,0.001227002,0.42258832,0.0016674881,0.006404327,0.0020207537,0.0034323304,0.07685254,0.07673702,0.08863447,0.005843297,0.30812582],"study_design_scores_gemma":[0.0005043234,0.0042830957,0.24457498,0.0002157412,0.00079413806,0.0021098035,0.002116934,0.48055124,0.06325577,0.19528152,0.0058391425,0.00047327756],"about_ca_topic_score_codex":0.0010939466,"about_ca_topic_score_gemma":0.0007163501,"teacher_disagreement_score":0.030461183,"about_ca_system_score_codex":0.0010301934,"about_ca_system_score_gemma":0.0018199655,"threshold_uncertainty_score":0.16109604},"labels":[],"label_agreement":null},{"id":"W4415333191","doi":"10.1093/biostatistics/kxaf033","title":"Wastewater-based reproduction rates for epidemic curve reconstruction","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Guelph; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bayesian probability; Inference; Autoregressive model; Bayesian inference; Key (lock); Poisson distribution; Estimator; Gaussian","score_opus":0.25291431894758687,"score_gpt":0.4580066023355928,"score_spread":0.20509228338800595,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415333191","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004687782,0.000068530484,0.99411726,0.00014616406,0.000009808907,0.000024940278,0.00017068993,0.00016233105,0.0006125345],"genre_scores_gemma":[0.43327314,0.0007873844,0.5578541,0.0002740139,0.00013546916,0.0003750782,0.001856324,0.00041404905,0.0050304416],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99813145,0.00091515266,0.00012176642,0.0003552302,0.00033001392,0.00014641596],"domain_scores_gemma":[0.9870099,0.008725479,0.0015225031,0.0014256333,0.0009791005,0.0003374232],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0077795754,0.00078236224,0.001109973,0.00273115,0.0005604396,0.001711155,0.0034091205,0.001745212,0.0046165744],"category_scores_gemma":[0.042472806,0.0009037438,0.001683497,0.0017736016,0.001693999,0.0029943646,0.0022811703,0.0027631011,0.00096979213],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000045944045,0.000027945409,0.006551752,0.000072971,0.000055431836,0.00014183285,0.00022780237,0.75348425,0.0011791162,0.19868293,0.0015274731,0.038002566],"study_design_scores_gemma":[0.000006744424,0.0000124933185,0.00056841236,0.000019919536,0.000008278498,0.00004719176,0.000020946689,0.9395216,0.00037236747,0.05818383,0.0012167799,0.000021390975],"about_ca_topic_score_codex":0.015011793,"about_ca_topic_score_gemma":0.010311501,"teacher_disagreement_score":0.015011793,"about_ca_system_score_codex":0.0018221341,"about_ca_system_score_gemma":0.0017722533,"threshold_uncertainty_score":0.04114276},"labels":[],"label_agreement":null},{"id":"W4415615025","doi":"10.1093/biostatistics/kxaf032","title":"A novel high-dimensional model for identifying regional DNA methylation QTLs","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre; McGill University; Université du Québec à Montréal; Mila - Quebec Artificial Intelligence Institute; Jewish General Hospital; York University","funders":"Canadian Statistical Sciences Institute; Natural Sciences and Engineering Research Council of Canada","keywords":"Interpretability; Covariate; Flexibility (engineering); Smoothness; Selection (genetic algorithm); Regression; Feature selection; Variable (mathematics); Regression analysis","score_opus":0.2821465805369406,"score_gpt":0.4362327932265029,"score_spread":0.1540862126895623,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415615025","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011108566,0.00012011337,0.9870624,0.0004200717,0.000028227047,0.000038448758,0.0003756569,0.0002006604,0.00064576906],"genre_scores_gemma":[0.48346904,0.00067307486,0.49593836,0.00069630804,0.00021075577,0.000989683,0.0021050095,0.00028977098,0.015628029],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986304,0.00070308155,0.000043048723,0.00034072323,0.00016406445,0.00011869402],"domain_scores_gemma":[0.99694437,0.0020736214,0.00031929606,0.00022117434,0.00031677412,0.00012479843],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0035101403,0.0009488577,0.0014656731,0.0009479315,0.0005801763,0.0012833067,0.0036373448,0.0021308386,0.0044247634],"category_scores_gemma":[0.008657959,0.00088154443,0.001533079,0.0013969183,0.0013977675,0.001489216,0.0019548202,0.0030215988,0.00084840367],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012879628,0.000055359047,0.0036146017,0.00009085383,0.00009296527,0.00018330483,0.0001148754,0.9229075,0.0015673265,0.049953565,0.00234973,0.018941106],"study_design_scores_gemma":[0.000018870012,0.00001853727,0.00032630604,0.000006942903,0.0000113512915,0.000029558132,0.0000075848166,0.9886362,0.00012541277,0.010191786,0.00061506464,0.000012334442],"about_ca_topic_score_codex":0.009995785,"about_ca_topic_score_gemma":0.009202477,"teacher_disagreement_score":0.009995785,"about_ca_system_score_codex":0.00092676666,"about_ca_system_score_gemma":0.001665533,"threshold_uncertainty_score":0.019875228},"labels":[],"label_agreement":null},{"id":"W4416052855","doi":"10.1093/biostatistics/kxaf034","title":"Markov switching zero-inflated space-time multinomial models for comparing multiple infectious diseases","year":2024,"lang":"en","type":"article","venue":"Biostatistics","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University Health Centre","funders":"Canada First Research Excellence Fund; Fundação Oswaldo Cruz; Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données; Escola Nacional de Saúde Pública Sérgio Arouca; Compute Canada","keywords":"Markov chain Monte Carlo; Markov chain; Multinomial distribution; Bayesian probability; Univariate; Inference; Bayesian inference; Autoregressive model; Multivariate statistics","score_opus":0.13757555115086603,"score_gpt":0.38399366470267043,"score_spread":0.2464181135518044,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416052855","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07037346,0.00081994996,0.9233927,0.00075130584,0.00041219292,0.0005468726,0.0015370456,0.00080373685,0.001362729],"genre_scores_gemma":[0.6834431,0.00085063314,0.30279034,0.00074478675,0.00047126098,0.0036729064,0.0035348448,0.00032682915,0.0041653076],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.92799795,0.0604903,0.0018290551,0.005922252,0.0022565792,0.0015038727],"domain_scores_gemma":[0.71272373,0.25312284,0.014167404,0.015155015,0.0030907234,0.0017403336],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.077585675,0.0021389113,0.003993281,0.005223608,0.0015137597,0.0033901588,0.0072886017,0.00428626,0.010863205],"category_scores_gemma":[0.18699183,0.0013649917,0.0064293086,0.0047183624,0.003916836,0.004704805,0.00410454,0.0061733313,0.0011120368],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025606016,0.000680902,0.07648454,0.00078196265,0.005315713,0.0012901523,0.0018082992,0.5916409,0.0015552215,0.2393027,0.0035605724,0.07501847],"study_design_scores_gemma":[0.00013152216,0.0005239437,0.006026683,0.00012270373,0.00031044622,0.00017846818,0.00018175747,0.8796212,0.00033015796,0.11050851,0.001976091,0.00008853346],"about_ca_topic_score_codex":0.009665039,"about_ca_topic_score_gemma":0.0057374043,"teacher_disagreement_score":0.077585675,"about_ca_system_score_codex":0.0032579526,"about_ca_system_score_gemma":0.002613975,"threshold_uncertainty_score":0.41031718},"labels":[],"label_agreement":null},{"id":"W939145883","doi":"10.1093/biostatistics/kxw049","title":"Inequality in treatment benefits: Can we determine if a new treatment benefits the many or the few?","year":2016,"lang":"en","type":"article","venue":"Biostatistics","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Neurological Disorders and Stroke; Patient-Centered Outcomes Research Institute; U.S. Food and Drug Administration; National Institute on Aging; National Institutes of Health; Hamilton Health Sciences Foundation","keywords":"Randomized controlled trial; Estimator; Fraction (chemistry); Computer science; Statistics; Econometrics; Randomized response; Outcome (game theory); Mathematics; Medicine; Mathematical optimization; Mathematical economics","score_opus":0.2502387195907997,"score_gpt":0.4054696846381312,"score_spread":0.15523096504733153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W939145883","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07654475,0.02712038,0.82765245,0.04618454,0.0010646032,0.0011349368,0.003986126,0.00032132966,0.015990864],"genre_scores_gemma":[0.8239372,0.0076049585,0.1443132,0.013600182,0.0018759731,0.002844637,0.001824615,0.00026545319,0.0037338366],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.93646604,0.0472337,0.0028319207,0.0072072865,0.0047854455,0.0014755272],"domain_scores_gemma":[0.5435908,0.42867422,0.014417306,0.009568579,0.0027460752,0.0010030228],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.11409813,0.0012558558,0.0076599745,0.0024032407,0.0010949831,0.0041931844,0.003377277,0.0044448143,0.012195695],"category_scores_gemma":[0.41495913,0.00090858364,0.0029176276,0.0030269686,0.0074649486,0.013041673,0.004020425,0.008993394,0.0009656073],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00426358,0.00048856856,0.04533226,0.0043198317,0.0033031339,0.00042271154,0.0017727371,0.022997214,0.0008384661,0.5008738,0.014272083,0.40111554],"study_design_scores_gemma":[0.0005145228,0.0003265785,0.010115388,0.0014830399,0.0009590246,0.00023183275,0.00021132138,0.03257346,0.00069212605,0.94427526,0.008552756,0.000064649576],"about_ca_topic_score_codex":0.0027730102,"about_ca_topic_score_gemma":0.0014602372,"teacher_disagreement_score":0.88590187,"about_ca_system_score_codex":0.0023679256,"about_ca_system_score_gemma":0.0026626468,"threshold_uncertainty_score":0.60341585},"labels":[],"label_agreement":null}]}