{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.018143,0.00008789195,0.0005089424,0.0004733698,0.000140504,0.0000139686,0.0001119668,0.00008444303,0.00004068432],"category_scores_gemma":[0.02418432,0.00006042782,0.00007172048,0.0008968541,0.0002833453,0.0001206009,0.00006376002,0.0003224973,0.00000200694],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001102212,"about_ca_system_score_gemma":0.0006179868,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001417966,"about_ca_topic_score_gemma":0.00005388646,"domain_scores_codex":[0.9958127,0.002681836,0.0004646299,0.0002302171,0.0004947395,0.0003159228],"domain_scores_gemma":[0.9825892,0.01544,0.0001480986,0.0004791762,0.001266349,0.00007715659],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0003979206,0.0005493408,0.1836292,0.002579404,0.001515335,0.000009448709,0.00874448,0.0001405094,0.0180557,0.003074845,0.0004687669,0.7808351],"study_design_scores_gemma":[0.00004535021,0.000162858,0.01345541,0.001719271,0.0003229978,0.000002037582,0.0679313,0.00535117,0.9049255,0.00584762,0.0001644611,0.000072019],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9262246,0.004547467,0.06261148,0.0007900948,0.00007790206,0.0004916782,0.000007053788,0.00001979974,0.005229955],"genre_scores_gemma":[0.8831037,0.001595302,0.114794,0.00001729831,0.00004412558,0.00007414991,0.000001317629,0.00001107353,0.0003590761],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8868698,"threshold_uncertainty_score":0.9840354,"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."}}