{"id":"W4412604178","doi":"10.1136/bjo-2025-327360","title":"Performance of DeepSeek-R1 in ophthalmology: an evaluation of clinical decision-making and cost-effectiveness","year":2025,"lang":"en","type":"article","venue":"British Journal of Ophthalmology","topic":"Retinal Imaging and Analysis","field":"Medicine","cited_by":17,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; St. Michael's Hospital; Hôpital Maisonneuve-Rosemont; Centre Hospitalier de l’Université de Montréal; McGill University; University of Ottawa; University of Waterloo; University of Toronto","funders":"","keywords":"McNemar's test; Medicine; Subspecialty; Ophthalmology; Optometry; Machine learning; Computer science; Statistics; Family medicine; Mathematics","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":[],"consensus_categories":[],"category_scores_codex":[0.007169979,0.00009840246,0.0009618048,0.000426617,0.00003334108,0.000008793154,0.0001405482,0.0001583502,0.0001116841],"category_scores_gemma":[0.003211101,0.0001008013,0.0001638519,0.000381838,0.0002732692,0.0001437915,0.00005047872,0.0003986786,3.765664e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004853743,"about_ca_system_score_gemma":0.0003329318,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007701833,"about_ca_topic_score_gemma":0.000001834046,"domain_scores_codex":[0.9970974,0.0009536773,0.001240146,0.0002062718,0.0003439243,0.0001586358],"domain_scores_gemma":[0.9971202,0.001015461,0.0006167644,0.0001562088,0.001020912,0.00007046277],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00107239,0.0003971429,0.8026662,0.0001308879,0.0001118336,0.002229705,0.00002277673,0.0004028653,0.0004456759,0.00000545704,0.00000761509,0.1925075],"study_design_scores_gemma":[0.002726552,0.001137802,0.8864761,0.002847124,0.0003547571,0.09788208,0.00009472844,0.007557783,0.0002012465,0.0006592117,0.00000342623,0.0000591952],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9954185,0.002506323,0.00009527715,0.00003785387,0.0002259916,0.0002113259,0.000002935635,0.000001680983,0.001500123],"genre_scores_gemma":[0.9978361,0.0002084956,0.001865767,0.00001259363,0.00004581864,0.000004929364,0.000002601562,0.000008181511,0.00001558755],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1924483,"threshold_uncertainty_score":0.4110558,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0785873835091986,"score_gpt":0.4703173014432211,"score_spread":0.3917299179340226,"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."}}