{"id":"W4399771702","doi":"10.2196/45973","title":"Performance Drift in Machine Learning Models for Cardiac Surgery Risk Prediction: Retrospective Analysis","year":2024,"lang":"en","type":"article","venue":"JMIRx Med","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"British Heart Foundation; University of Bristol; National Institute for Health and Care Research","keywords":"EuroSCORE; Concept drift; Medicine; Metric (unit); Cardiac surgery; Data set; Retrospective cohort study; Performance metric; Generalization; Set (abstract data type); Artificial intelligence; Machine learning; Computer science; Cardiology; Internal medicine; Engineering; Operations management; Data stream mining; Mathematics","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001058342,0.0001047224,0.000368863,0.0004917454,0.0001134886,0.00002488454,0.00003158438,0.0001060835,0.00006584801],"category_scores_gemma":[0.0003589706,0.00009346603,0.0002453712,0.001142151,0.00002791648,0.000174088,0.000008664634,0.0003731612,0.00001948374],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002521248,"about_ca_system_score_gemma":0.0001898973,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008155098,"about_ca_topic_score_gemma":0.0001599882,"domain_scores_codex":[0.9987857,0.00007089259,0.0003966429,0.0003062887,0.0002027426,0.0002377658],"domain_scores_gemma":[0.9991122,0.0004103767,0.00006016454,0.0001675531,0.0001601181,0.00008956193],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001059099,0.00003850097,0.9541011,0.000176395,0.0002639356,0.00000310066,0.002688862,0.003823416,0.00003398755,0.00005871874,0.002185375,0.03652066],"study_design_scores_gemma":[0.00002304685,0.0001359899,0.1506061,0.0001369863,0.0005012497,0.000001859783,0.000533806,0.8411253,0.0008176378,0.0004879625,0.00553045,0.00009959406],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9906449,0.001380199,0.003333857,0.002031843,0.001056528,0.000595975,0.00005335449,0.0001521607,0.000751227],"genre_scores_gemma":[0.9948705,0.00139351,0.0001598089,0.00003922381,0.0005673629,0.0002638908,0.0002176149,0.00001936002,0.002468725],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8373019,"threshold_uncertainty_score":0.3811434,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1038413284884274,"score_gpt":0.3741849249451892,"score_spread":0.2703435964567619,"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."}}