{"id":"W6939181405","doi":"10.60692/cbrxw-n5f07","title":"EHR Foundation Models Improve Robustness in the Presence of Temporal Distribution Shift","year":2022,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Institute for Clinical Evaluative Sciences; SickKids Foundation; Hospital for Sick Children","funders":"","keywords":"Robustness (evolution); Logistic regression; Transformer; Health records; Regression; Electronic health record; Regression analysis; Predictive modelling","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":[0.001577024,0.000107762,0.0001469191,0.0001499742,0.0002584698,0.0001627376,0.0008824581,0.00004018093,0.000005104809],"category_scores_gemma":[0.00003205399,0.00008415116,0.00004320094,0.0006120178,0.000019557,0.001573151,0.0002782001,0.0002380874,0.00001300722],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001980753,"about_ca_system_score_gemma":0.00009121079,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001998101,"about_ca_topic_score_gemma":0.000001357401,"domain_scores_codex":[0.99791,0.0004616673,0.0006344908,0.0001588527,0.0006357209,0.0001992936],"domain_scores_gemma":[0.9987604,0.0000304658,0.0004809639,0.000582697,0.0001150214,0.00003044432],"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":[0.00004517417,0.000008435009,0.1525576,0.0006786321,0.000008668525,0.000003684682,0.2323914,0.5556485,3.223198e-7,0.05604192,0.00005613667,0.002559591],"study_design_scores_gemma":[0.0003095006,0.00006254308,0.03222033,0.00003052439,0.000002271844,0.00001852967,0.006097649,0.9610047,0.000009319058,0.00004760794,0.0001017773,0.00009520673],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2847797,0.000001797305,0.7134872,0.0004153398,0.0003783396,0.0005007639,0.00005748474,0.00008957428,0.0002897622],"genre_scores_gemma":[0.9989101,2.484513e-8,0.0006901061,0.00007060642,0.00002225263,0.0002094617,0.0000835188,0.000003323859,0.00001058383],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7141304,"threshold_uncertainty_score":0.3431585,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03510548115316262,"score_gpt":0.2448333130762237,"score_spread":0.209727831923061,"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."}}