{"id":"W4400589155","doi":"10.1007/s00330-024-10902-5","title":"ChatGPT’s diagnostic performance based on textual vs. visual information compared to radiologists’ diagnostic performance in musculoskeletal radiology","year":2024,"lang":"en","type":"article","venue":"European Radiology","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":57,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Guerbet","keywords":"Medicine; Neuroradiology; Interventional radiology; Diagnostic accuracy; Radiology; Medical diagnosis; Neurology","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":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001201296,0.00034728,0.0005508016,0.0008142912,0.0001507161,0.00004275438,0.0002518978,0.0001802098,0.0001909109],"category_scores_gemma":[0.003526488,0.000305417,0.000112767,0.0006385513,0.0002581505,0.0002841925,0.00005427847,0.0007329641,0.003549671],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003985146,"about_ca_system_score_gemma":0.0003029134,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008954239,"about_ca_topic_score_gemma":0.00001408872,"domain_scores_codex":[0.9969548,0.000589492,0.0009571735,0.000537007,0.0002168722,0.0007446795],"domain_scores_gemma":[0.9961844,0.002898977,0.0001046237,0.000409555,0.0001099156,0.0002925226],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.002078026,0.0005492472,0.5045852,0.001719648,0.00009077155,0.0006629693,0.01015925,0.07816339,0.0003167739,0.0007095161,0.01949935,0.3814659],"study_design_scores_gemma":[0.0004845208,0.007228124,0.7681444,0.0009634658,0.00006218905,0.0006431976,0.0007177489,0.1813723,0.0004072668,0.00001424432,0.03939534,0.0005671265],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9874731,0.0004269101,0.00103823,0.003269459,0.002043163,0.0009969721,0.00001183695,0.000242986,0.004497302],"genre_scores_gemma":[0.994525,0.0005112663,0.0001857827,0.003031982,0.001073301,0.0001314321,0.0004038034,0.00004547524,0.00009200733],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3808987,"threshold_uncertainty_score":0.9999398,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04325767552305441,"score_gpt":0.3517228958222935,"score_spread":0.3084652202992391,"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."}}