{"id":"W4415020931","doi":"10.1017/rsm.2025.10034","title":"Using large language models to directly screen electronic databases as an alternative to traditional search strategies such as the Cochrane highly sensitive search for filtering randomized controlled trials in systematic reviews","year":2025,"lang":"en","type":"article","venue":"Research Synthesis Methods","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Cochrane; Hotel Dieu Hospital","funders":"","keywords":"Systematic review; Randomized controlled trial; MEDLINE; Keyword search; Cochrane collaboration; Electronic database; Filter (signal processing); Meta-analysis","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","metaepi_broad","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.8866688,0.0006330807,0.02314781,0.003317047,0.0007141744,0.00258811,0.002958972,0.0001252539,0.0009043589],"category_scores_gemma":[0.6872116,0.0002736107,0.004067821,0.004361481,0.000206839,0.0007458497,0.0004691576,0.0006894552,0.0002077428],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004571832,"about_ca_system_score_gemma":0.001393638,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001644637,"about_ca_topic_score_gemma":0.0004422978,"domain_scores_codex":[0.3321342,0.6387397,0.01590548,0.002210174,0.009270117,0.001740286],"domain_scores_gemma":[0.1647188,0.8220882,0.002836174,0.00502249,0.004730766,0.0006035049],"domain_codex":"methods","domain_gemma":"methods","domain_candidate":"methods","domain_consensus":"methods","study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.07920998,0.0009489641,0.000009048385,0.01429474,0.01089765,0.00009354755,0.02859735,0.02973774,0.08975711,0.7136908,0.002063225,0.03069985],"study_design_scores_gemma":[0.0316384,0.0003399039,0.0000149343,0.01651311,0.002382859,0.00002797952,0.04782804,0.8070372,0.04967645,0.0429328,0.00078627,0.0008220609],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05166774,0.00634182,0.9002811,0.001846646,0.0001135354,0.03591698,0.0003007305,0.00001426702,0.00351719],"genre_scores_gemma":[0.7065411,0.001079156,0.2704123,0.0005987975,0.0003669809,0.01692187,0.00004714185,0.00007969214,0.003952974],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7772995,"threshold_uncertainty_score":0.9999716,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.8796393867588007,"score_gpt":0.6853275020424803,"score_spread":0.1943118847163204,"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."}}