{"id":"W4410340882","doi":"10.1007/s43441-025-00785-z","title":"Using Large Language Models for Advanced and Flexible Labelling of Protocol Deviations in Clinical Development","year":2025,"lang":"en","type":"article","venue":"Therapeutic Innovation & Regulatory Science","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Roche (Canada)","funders":"","keywords":"Context (archaeology); Protocol (science); Computer science; Artificial intelligence; Natural language processing; Medicine; Pathology; Geography; Alternative medicine","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":[0.05775068,0.002875362,0.00144071,0.004701408,0.001712168,0.0097273,0.004829831,0.002921799,0.009632024],"category_scores_gemma":[0.182325,0.002049025,0.005460585,0.00262028,0.001971948,0.007474713,0.008031983,0.007169723,0.009311886],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005130807,"about_ca_system_score_gemma":0.0124567,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006688624,"about_ca_topic_score_gemma":0.01929096,"domain_scores_codex":[0.9347838,0.04098189,0.007835814,0.008712738,0.006779049,0.0009067283],"domain_scores_gemma":[0.8094662,0.1430781,0.01258232,0.01761474,0.0155716,0.001687036],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001588011,0.001039983,0.01377354,0.005624683,0.001384788,0.001062718,0.007736939,0.08661518,0.01118136,0.07753032,0.09356486,0.6988976],"study_design_scores_gemma":[0.000445302,0.0004182669,0.002762114,0.002044786,0.0005633223,0.0006887461,0.0008919457,0.6577765,0.01326564,0.1606947,0.1600382,0.0004104107],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00385836,0.0004730716,0.9599136,0.001938205,0.0002824242,0.002087884,0.004565909,0.02453787,0.00234275],"genre_scores_gemma":[0.02754886,0.0002514079,0.9591794,0.0008017342,0.0000874205,0.002692365,0.006978344,0.001567212,0.0008932842],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9422493,"threshold_uncertainty_score":0.3054185,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09952621716643238,"score_gpt":0.4547115080606196,"score_spread":0.3551852908941872,"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."}}