{"id":"W3197873762","doi":"10.1038/s41598-022-07167-7","title":"A comparison of approaches to improve worst-case predictive model performance over patient subpopulations","year":2022,"lang":"en","type":"preprint","venue":"Scientific Reports","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"National Heart, Lung, and Blood Institute; Stanford Research Computing Center, Stanford University; National Science Foundation","keywords":"Metric (unit); Computer science; Machine learning; Performance metric; Scale (ratio); Population; Predictive modelling; Predictive power; Quality (philosophy); Empirical research; Artificial intelligence; Medicine; Statistics; Mathematics; Operations management","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03865131,0.00186709,0.001706005,0.002180825,0.0007404985,0.001934972,0.002400748,0.00173601,0.000842964],"category_scores_gemma":[0.08958647,0.0006311986,0.001930551,0.001618272,0.001265398,0.003853335,0.003123202,0.00308391,0.0002550251],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001664725,"about_ca_system_score_gemma":0.002618761,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005638232,"about_ca_topic_score_gemma":0.006185837,"domain_scores_codex":[0.9844797,0.01029017,0.001127516,0.002064139,0.001626472,0.0004120046],"domain_scores_gemma":[0.9201113,0.06140121,0.00292847,0.009794766,0.004967603,0.0007966768],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007836635,0.0003911429,0.0194119,0.0002457448,0.0009418808,0.0000522467,0.0002592473,0.8629097,0.001020511,0.005649293,0.001508936,0.1068257],"study_design_scores_gemma":[0.00006132505,0.0004605199,0.00371564,0.00004817191,0.0001128398,0.00005189705,0.0001021002,0.9862882,0.001725778,0.006769701,0.000626433,0.00003739488],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.303921,0.004085874,0.6830471,0.002215164,0.0001414435,0.0004700603,0.0006502745,0.001289085,0.004179998],"genre_scores_gemma":[0.7953438,0.0008255565,0.2009203,0.0004160724,0.00008493683,0.0002767661,0.001306721,0.0001843631,0.0006416314],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03865131,"threshold_uncertainty_score":0.2044101,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1232113807884043,"score_gpt":0.3371047260080542,"score_spread":0.2138933452196499,"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."}}