{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002785763,0.000161895,0.0009050403,0.000228813,0.00006208109,0.00008880284,0.0003086043,0.0003209541,0.002595616],"category_scores_gemma":[0.07489123,0.0001288804,0.0003146757,0.0004316614,0.0005693031,0.0001302955,0.0001078748,0.00119783,0.00001221978],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000721797,"about_ca_system_score_gemma":0.0007671515,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004080449,"about_ca_topic_score_gemma":0.000003175552,"domain_scores_codex":[0.9969342,0.0004887254,0.001323147,0.0002630856,0.0005983206,0.0003925754],"domain_scores_gemma":[0.9794284,0.01949932,0.0003359476,0.0001984362,0.0002599555,0.0002779799],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"case_report","study_design_gemma":"case_report","study_design_scores_codex":[0.0001146708,0.0008897771,0.1493557,0.0001229453,0.0002136157,0.8118492,0.0001427788,0.00003243112,0.0001647839,0.0003287285,0.002518486,0.03426696],"study_design_scores_gemma":[0.0007900648,0.0005626087,0.3572864,0.006016602,0.00008051741,0.6341794,0.000157014,0.0001879054,0.00003227792,0.0004615137,0.0001736469,0.00007205766],"study_design_candidate":"case_report","study_design_consensus":"case_report","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9813557,0.009875014,0.00004713369,0.00191829,0.0008597608,0.0001257493,0.000008736615,0.00001177027,0.005797818],"genre_scores_gemma":[0.9975375,0.0003400838,0.001228897,0.0001721979,0.0004779573,0.00000419162,0.00001168953,0.00002984481,0.0001976209],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2079307,"threshold_uncertainty_score":0.9983162,"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."}}