{"id":"W4404727371","doi":"10.1177/17407745241297947","title":"Detecting irregularities in randomized controlled trials using machine learning","year":2024,"lang":"en","type":"article","venue":"Clinical Trials","topic":"Statistical Methods in Clinical Trials","field":"Mathematics","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Impact; Hamilton Health Sciences; McMaster University; Population Health Research Institute; University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Interquartile range; Artificial intelligence; Machine learning; Concordance; Outlier; Clinical trial; Computer science; Randomized controlled trial; Receiver operating characteristic; Medicine; Anomaly detection; Algorithm; Statistics; Data mining; Mathematics; Surgery; Pathology","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.7168379,0.00064738,0.03249818,0.0005430378,0.0001609418,0.0005187495,0.0004538014,0.0009738379,0.001673604],"category_scores_gemma":[0.9900886,0.0003986767,0.008414073,0.0006554119,0.0006888594,0.0001618418,0.0002326622,0.002408985,0.00006542906],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001174898,"about_ca_system_score_gemma":0.0003442875,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006119238,"about_ca_topic_score_gemma":0.00001466525,"domain_scores_codex":[0.6120554,0.3420426,0.04159069,0.001800099,0.001355688,0.001155586],"domain_scores_gemma":[0.02832595,0.9664767,0.004007876,0.0006431999,0.000209003,0.0003372425],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"randomized_trial","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.484295,0.0009571963,0.0005704246,0.001437132,0.009303787,0.0005035646,0.0004334965,0.000170771,0.001379014,0.3082657,0.0007266857,0.1919572],"study_design_scores_gemma":[0.2671473,0.0001284661,0.000004333162,0.001028418,0.002719359,0.000007768889,0.00006540101,0.05848923,0.0002168664,0.6694762,0.0003742016,0.0003424447],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1077571,0.02206917,0.8045095,0.002803313,0.02900927,0.02701296,0.0002598356,0.0021176,0.00446134],"genre_scores_gemma":[0.1660697,0.001777652,0.8216071,0.0003322949,0.007692627,0.0007849278,0.000006975747,0.0002487778,0.001479915],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.6244342,"threshold_uncertainty_score":0.9998925,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.8686945249905104,"score_gpt":0.6957301326060387,"score_spread":0.1729643923844717,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}