{"id":"W4391878439","doi":"10.1136/bjo-2023-325053","title":"Assessing the medical reasoning skills of GPT-4 in complex ophthalmology cases","year":2024,"lang":"en","type":"article","venue":"British Journal of Ophthalmology","topic":"Clinical Reasoning and Diagnostic Skills","field":"Medicine","cited_by":54,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; University of Waterloo; McGill University; Centre Hospitalier de l’Université de Montréal; Institut Universitaire en Santé Mentale de Québec; Université de Montréal; Hôpital du Sacré-Cœur de Montréal; Institut universitaire en santé mentale de Montréal; Hôpital Maisonneuve-Rosemont","funders":"","keywords":"Medicine; Optometry; Ophthalmology","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":[],"consensus_categories":[],"category_scores_codex":[0.005835201,0.000736077,0.0002793829,0.000679282,0.0002149783,0.001564836,0.001328779,0.001191121,0.00458302],"category_scores_gemma":[0.03474249,0.0003555608,0.0006134856,0.000255751,0.0008143379,0.001363418,0.001876466,0.00135768,0.001860037],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001104244,"about_ca_system_score_gemma":0.001293801,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002371023,"about_ca_topic_score_gemma":0.002024108,"domain_scores_codex":[0.9974054,0.001196323,0.0002065543,0.0005649255,0.0004120939,0.0002147206],"domain_scores_gemma":[0.982906,0.0113881,0.001313332,0.00145591,0.001917433,0.001019231],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001598072,0.001742415,0.6816917,0.0007004113,0.000254228,0.001066453,0.008998433,0.05716958,0.01663212,0.001450956,0.009546598,0.2191491],"study_design_scores_gemma":[0.000514698,0.005932664,0.3055476,0.0007065601,0.0005445609,0.004889095,0.00832949,0.5878606,0.04391395,0.01018342,0.03132555,0.0002517403],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9760418,0.0001772998,0.01801602,0.0004784979,0.00003976632,0.0001825499,0.0004494251,0.0009334088,0.003681156],"genre_scores_gemma":[0.9864274,0.00009957417,0.01149877,0.0001634416,0.000007746171,0.00006564151,0.0009061962,0.00005041878,0.0007808217],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005835201,"threshold_uncertainty_score":0.03085989,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06008550344862287,"score_gpt":0.4240634329352387,"score_spread":0.3639779294866159,"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."}}