{"id":"W4379769651","doi":"10.1038/s41586-023-06160-y","title":"Health system-scale language models are all-purpose prediction engines","year":2023,"lang":"en","type":"article","venue":"Nature","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":447,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadian Institute for Advanced Research","funders":"National Cancer Institute; W. M. Keck Foundation","keywords":"Generalizability theory; Software deployment; Computer science; Machine learning; Predictive modelling; Artificial intelligence; Clinical decision support system; Language model; Scale (ratio); Health care; Data science; Decision support system; Psychology; 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":[],"consensus_categories":[],"category_scores_codex":[0.0005799739,0.0001454964,0.0002140758,0.0001978238,0.0001614084,0.00007608373,0.0006343561,0.0005580824,0.000003186505],"category_scores_gemma":[0.00006429136,0.0001306843,0.00005847369,0.0007952392,0.000009896261,0.0002891308,0.0001838464,0.001828308,0.00008515103],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001539561,"about_ca_system_score_gemma":0.00009451824,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001995849,"about_ca_topic_score_gemma":0.00009749623,"domain_scores_codex":[0.9982787,0.0001779796,0.0002176902,0.0004330994,0.0004535832,0.0004389547],"domain_scores_gemma":[0.9988975,0.00007759644,0.0001282075,0.0006313432,0.00008783361,0.00017753],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00004093917,0.0001645431,0.02886937,0.00695798,0.0001192342,0.000555051,0.04214183,0.1715059,0.000426466,0.104964,0.525782,0.1184727],"study_design_scores_gemma":[0.0001849936,0.0000611541,0.01201603,0.0002216252,0.000002788806,0.00004925592,0.0003877279,0.9773848,0.00003092404,0.0002769453,0.009239449,0.000144314],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2762909,0.02946404,0.3297453,0.2945341,0.02117044,0.004440148,0.0006998497,0.03498203,0.008673154],"genre_scores_gemma":[0.9905565,0.00005695625,0.005833549,0.002132382,0.0004084425,0.00003809129,0.00006858807,0.00002514215,0.0008803967],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8058789,"threshold_uncertainty_score":0.7943192,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01301851991908903,"score_gpt":0.2999892234336552,"score_spread":0.2869707035145662,"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."}}