{"id":"W4384499027","doi":"10.1097/icu.0000000000000986","title":"Toward safer ophthalmic artificial intelligence via distributed validation on real-world data","year":2023,"lang":"en","type":"review","venue":"Current Opinion in Ophthalmology","topic":"Retinal Imaging and Analysis","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Context (archaeology); Machine learning; SAFER; Computer science; Implementation; Process (computing); Medicine; Artificial intelligence; Pipeline (software); Data science; Risk analysis (engineering); Computer security; Software engineering","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":[0.0009606925,0.000639165,0.002511767,0.001413768,0.00007426376,0.0000538068,0.0008720804,0.0002722693,0.0002865854],"category_scores_gemma":[0.0005914785,0.000544821,0.0004462079,0.002486001,0.0002291242,0.00009107011,0.0004878287,0.001303874,0.001326951],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003744237,"about_ca_system_score_gemma":0.0003421428,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002200897,"about_ca_topic_score_gemma":0.000001803317,"domain_scores_codex":[0.9952104,0.000615293,0.001688338,0.001423544,0.0004662488,0.0005961774],"domain_scores_gemma":[0.9966663,0.0006611918,0.0006930712,0.001668367,0.0001157166,0.0001953992],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00009986412,0.0006047674,0.0007060253,0.02247584,0.0002668058,0.0003627139,0.00002800053,0.0000250395,3.240134e-7,0.0007016012,0.002190599,0.9725384],"study_design_scores_gemma":[0.000370022,0.0008158,0.0009802718,0.08888482,0.002243984,0.004842465,0.00008421402,0.003361411,0.000007082589,0.003216182,0.8933753,0.001818395],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.0001356776,0.9848571,0.001049129,0.0006129618,0.009785895,0.001546455,0.001297373,0.0001925705,0.0005228229],"genre_scores_gemma":[0.000806358,0.9566621,0.0000822625,0.00000386557,0.00132602,0.0001368047,0.04069556,0.00009590883,0.0001911429],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9707201,"threshold_uncertainty_score":0.9997003,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.5447142235765434,"score_gpt":0.5332955114215067,"score_spread":0.01141871215503676,"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."}}