{"id":"W4413422599","doi":"10.1016/j.acra.2025.08.014","title":"Evaluating Large Language Models for Radiology Systematic Review Title and Abstract Screening","year":2025,"lang":"en","type":"article","venue":"Academic Radiology","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of British Columbia","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institutes of Health; American Roentgen Ray Society; U.S. Department of Defense","keywords":"Computer science; Medical physics; Medicine; Radiology","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":[],"category_scores_codex":[0.0931176,0.005495447,0.01457931,0.01197175,0.001578669,0.005608193,0.004626823,0.003546192,0.01214125],"category_scores_gemma":[0.3082424,0.003391348,0.04170524,0.007394661,0.001480868,0.005322692,0.004084319,0.00313469,0.002127619],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003553517,"about_ca_system_score_gemma":0.01058529,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008767117,"about_ca_topic_score_gemma":0.03176315,"domain_scores_codex":[0.8945848,0.07787973,0.01708122,0.005936829,0.004020695,0.0004968964],"domain_scores_gemma":[0.6186301,0.3475772,0.01840001,0.008249187,0.005831792,0.001311675],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.01232238,0.0002285037,0.01659582,0.4414241,0.4251485,0.0005883543,0.0006389609,0.01897035,0.001670219,0.001102547,0.01476545,0.06654475],"study_design_scores_gemma":[0.01112687,0.002822352,0.01122756,0.05955296,0.8524867,0.0007468256,0.0004159859,0.03547652,0.0008848625,0.00920844,0.01561001,0.0004409929],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.06603226,0.7299799,0.07337104,0.008752049,0.003156411,0.02050324,0.08885577,0.005502037,0.003847345],"genre_scores_gemma":[0.6742305,0.1111635,0.1205675,0.008375011,0.001857299,0.03614932,0.0433401,0.001578205,0.002738622],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9068824,"threshold_uncertainty_score":0.4924588,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.243666170429819,"score_gpt":0.5292139635232682,"score_spread":0.2855477930934492,"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."}}