{"id":"W4409260121","doi":"10.52225/narra.v5i1.2371","title":"Chinese generative AI models (DeepSeek and Qwen) rival ChatGPT-4 in ophthalmology queries with excellent performance in Arabic and English","year":2025,"lang":"en","type":"article","venue":"Narra J","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":20,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Arabic; Generative grammar; Computer science; Natural language processing; Linguistics; Artificial intelligence; Ophthalmology; Medicine; Philosophy","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":[0.000185176,0.0001107581,0.0002245604,0.0001884468,0.00006046992,0.00001274526,0.00003110708,0.00009385424,0.00001818603],"category_scores_gemma":[0.00008731867,0.00008598351,0.000009395903,0.00025754,0.000123391,0.0001821749,0.00002116069,0.0002425757,0.000001149501],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006595707,"about_ca_system_score_gemma":0.0001783638,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006486773,"about_ca_topic_score_gemma":0.000568273,"domain_scores_codex":[0.9992221,0.00004591079,0.0002332883,0.0002352854,0.00007676304,0.0001866727],"domain_scores_gemma":[0.999622,0.00006727373,0.00003187541,0.0001180691,0.0001041752,0.00005656596],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0005153041,0.0001289379,0.9457862,0.0002235003,0.00001067339,0.00003769121,0.04276972,0.0006458833,0.0001504077,0.0005183523,0.00004720157,0.009166069],"study_design_scores_gemma":[0.0009993872,0.001789914,0.8892102,0.0009800617,0.00003116179,0.0002506743,0.02201219,0.06885573,0.004568024,0.01039291,0.0004807467,0.0004289976],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9936427,0.001280714,0.00007458864,0.00290753,0.0002094925,0.0003355742,8.74512e-7,0.00001202285,0.001536535],"genre_scores_gemma":[0.9982501,0.0004562138,0.0002234365,0.0004507506,0.0001009507,0.00007000849,0.000006573788,0.000006956639,0.0004349712],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.06820984,"threshold_uncertainty_score":0.3506306,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04285039109542241,"score_gpt":0.3732398101649944,"score_spread":0.330389419069572,"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."}}