{"id":"W4412513187","doi":"10.3390/diagnostics15141820","title":"Evaluating ChatGPT-4 Plus in Ophthalmology: Effect of Image Recognition and Domain-Specific Pretraining on Diagnostic Performance","year":2025,"lang":"en","type":"article","venue":"Diagnostics","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Sherbrooke","funders":"Université de Sherbrooke","keywords":"Artificial intelligence; Ophthalmology; Domain (mathematical analysis); Computer science; Image (mathematics); Psychology; Optometry; Pattern recognition (psychology); Medicine; Computer vision; Mathematics","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.0008747583,0.0001268702,0.0002925453,0.0002459513,0.00007016866,0.00001113366,0.00004612649,0.0001225469,0.00004113032],"category_scores_gemma":[0.005664197,0.0001181174,0.00003002348,0.000314184,0.0001150125,0.00005854883,0.00002272552,0.0002638318,0.00002906706],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001250638,"about_ca_system_score_gemma":0.0001176586,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006914571,"about_ca_topic_score_gemma":0.000003427488,"domain_scores_codex":[0.9988372,0.0001381665,0.0004123029,0.0002394142,0.0001526826,0.0002202827],"domain_scores_gemma":[0.9916414,0.00790196,0.00009399318,0.0001773298,0.0001291335,0.00005615534],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0006623673,0.0001934656,0.4621299,0.001348506,0.00002259271,0.00007646987,0.004875035,0.0001354443,0.001149584,0.0001739525,0.0002483214,0.5289844],"study_design_scores_gemma":[0.001883952,0.01201966,0.7717937,0.01411042,0.0002355468,0.0001662135,0.003588816,0.008056339,0.1813559,0.005912765,0.0003384793,0.0005382553],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9957871,0.001088174,0.00009299381,0.0004054127,0.0004531894,0.0007655151,0.000006801138,0.00001707023,0.001383672],"genre_scores_gemma":[0.9965774,0.002001341,0.0009910739,0.0001181946,0.0001089664,0.0001215362,0.00005073417,0.00001176248,0.00001893385],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5284461,"threshold_uncertainty_score":0.6780983,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1817417859717197,"score_gpt":0.456601610044809,"score_spread":0.2748598240730893,"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."}}