{"id":"W4415396422","doi":"10.1002/cesm.70042","title":"Enhancing Evidence Synthesis Efficiency: Leveraging Large Language Models and Agentic Workflows for Optimized Literature Screening","year":2025,"lang":"en","type":"article","venue":"Cochrane Evidence Synthesis and Methods","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Public Health Agency of Canada; University of Waterloo; Health Canada","funders":"Public Health Agency; World Health Organization","keywords":"Workflow; Public health; Key (lock); Language model; Health care; Component (thermodynamics)","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","metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.006429962,0.0002928088,0.0007385711,0.0003406241,0.0005190872,0.0002096936,0.0001647693,0.0002206849,0.00005300937],"category_scores_gemma":[0.02672588,0.0002532258,0.0001646916,0.0005671392,0.0001145614,0.0005690962,0.00009382825,0.0002905027,0.000001567896],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000753512,"about_ca_system_score_gemma":0.0002372777,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001264012,"about_ca_topic_score_gemma":0.00001386044,"domain_scores_codex":[0.9972134,0.0006107614,0.0006565761,0.00072473,0.0002463663,0.0005481141],"domain_scores_gemma":[0.9799603,0.0188477,0.0001616599,0.000476598,0.0003306679,0.0002230765],"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":[0.0007494178,0.00009460218,0.0008909102,0.005021789,0.0001105135,0.00001847332,0.0114924,0.0002367225,0.05916191,0.0004795197,0.0001359341,0.9216078],"study_design_scores_gemma":[0.0002126347,0.0001588116,0.001081475,0.09033535,0.001549044,0.00007104934,0.01078985,0.2016914,0.6909218,0.002244539,0.0003061041,0.000637927],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1263974,0.1127349,0.7555528,0.003985261,0.0003295258,0.0008402515,0.000004534809,0.00007551254,0.00007981655],"genre_scores_gemma":[0.649205,0.01184633,0.3373814,0.0006193173,0.0001776834,0.0003455825,0.000001869267,0.00002429594,0.0003984683],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9209699,"threshold_uncertainty_score":0.999992,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1271447016142044,"score_gpt":0.4821077098525858,"score_spread":0.3549630082383814,"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."}}