{"id":"W4413135087","doi":"10.1017/rsm.2025.10028","title":"Assessing risk of bias of cohort studies with large language models","year":2025,"lang":"en","type":"article","venue":"Research Synthesis Methods","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"McMaster University","funders":"","keywords":"Cohort; Consistency (knowledge bases); Kappa; Confidence interval; Population; Cohort study; Medicine; Gold standard (test); Statistics; Demography; Psychology; Internal medicine; Environmental health; Computer science; Mathematics; Artificial intelligence","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.6809758,0.003392966,0.006043757,0.01130876,0.00204577,0.007187745,0.004405662,0.003764368,0.005005433],"category_scores_gemma":[0.8208266,0.002615782,0.01959193,0.005842451,0.00447754,0.006925777,0.009240885,0.003285896,0.0005189266],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003293898,"about_ca_system_score_gemma":0.005708678,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002259617,"about_ca_topic_score_gemma":0.003227207,"domain_scores_codex":[0.2297574,0.631427,0.0878381,0.01831438,0.03136391,0.001299165],"domain_scores_gemma":[0.05417231,0.858513,0.04554206,0.03085379,0.01042222,0.0004966414],"domain_codex":"methods","domain_gemma":"methods","domain_candidate":"methods","domain_consensus":"methods","study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.01218528,0.0004912789,0.4787852,0.0560838,0.1322433,0.001649078,0.01183084,0.02434446,0.002556022,0.02531616,0.005947495,0.2485671],"study_design_scores_gemma":[0.008222583,0.01283858,0.219403,0.05932875,0.1548855,0.006799662,0.005868588,0.300251,0.01582194,0.1583693,0.05599109,0.002219826],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1300454,0.0336471,0.8008748,0.004771562,0.001490052,0.01995534,0.002964,0.001178386,0.00507334],"genre_scores_gemma":[0.6843117,0.002160317,0.2845001,0.001407832,0.0003597394,0.02554736,0.0009753612,0.0001652482,0.0005722367],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.3190242,"threshold_uncertainty_score":0.3934137,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.6953569907842155,"score_gpt":0.6929927833746521,"score_spread":0.002364207409563446,"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."}}