{"id":"W7115709189","doi":"10.48448/gb1y-9631","title":"Understanding How a Language Model Assesses the Quality of Randomized Controlled Trials: Applying Shapley Additive Explanations to Encoder Transformer Classification Models","year":2025,"lang":"","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McMaster University","funders":"","keywords":"Language model; Workflow; Quality (philosophy); Test (biology); Critical appraisal; Deep learning","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"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":[0.1506418,0.001433832,0.002166313,0.006034223,0.001099551,0.007749744,0.00341806,0.003760993,0.008311112],"category_scores_gemma":[0.5026098,0.001005297,0.003689519,0.003173055,0.00348339,0.00610014,0.002908706,0.005085161,0.001064256],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005826789,"about_ca_system_score_gemma":0.007564905,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005186956,"about_ca_topic_score_gemma":0.00834842,"domain_scores_codex":[0.8765405,0.1020961,0.008032341,0.005824329,0.006693375,0.0008133723],"domain_scores_gemma":[0.3099941,0.6445693,0.02147913,0.01210082,0.01066736,0.001189343],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00355297,0.0002851001,0.05771865,0.008069746,0.003223262,0.0006976813,0.004642992,0.1375731,0.001284106,0.2421065,0.02859374,0.5122522],"study_design_scores_gemma":[0.0006310423,0.0003997843,0.005093159,0.002631988,0.001156174,0.0002284821,0.0004456698,0.5371405,0.002162189,0.4406314,0.009287202,0.0001922917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04540607,0.005041249,0.9041201,0.03133415,0.0006899037,0.001865535,0.004329117,0.001629326,0.005584674],"genre_scores_gemma":[0.6442972,0.001192945,0.3427572,0.005073236,0.0006045385,0.002340308,0.002034013,0.0001564395,0.001544236],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8493581,"threshold_uncertainty_score":0.7966797,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3430963393206387,"score_gpt":0.4309888300504897,"score_spread":0.087892490729851,"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."}}