{"id":"W4407093098","doi":"10.48550/arxiv.2501.18891","title":"CAAT-EHR: Cross-Attentional Autoregressive Transformer for Multimodal Electronic Health Record Embeddings","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institutes of Health Research; National Institutes of Health; Genentech; IXICO; H. Lundbeck A/S; Servier; Eisai; University of Southern California; Biogen; Eli Lilly and Company; Bristol-Myers Squibb; BioClinica; U.S. Department of Defense; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Pfizer; National Institute of General Medical Sciences; Northern California Institute for Research and Education; University of North Texas; Alzheimer's Association","keywords":"Autoregressive model; Transformer; Electronic health record; Health records; Computer science; Psychology; Econometrics; Engineering; Economics; Electrical engineering; Health care","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001269701,0.001431566,0.0006263112,0.0007136767,0.0002851993,0.0007521929,0.001479507,0.0009032755,0.003158932],"category_scores_gemma":[0.005486053,0.0004782348,0.001220571,0.0006741491,0.0004324696,0.001818039,0.001901625,0.002192499,0.001893575],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007535388,"about_ca_system_score_gemma":0.001279989,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007285203,"about_ca_topic_score_gemma":0.01262887,"domain_scores_codex":[0.9993981,0.0001720927,0.00003864791,0.0002005763,0.0001169365,0.00007357455],"domain_scores_gemma":[0.9990086,0.0004323725,0.00008397488,0.0001912852,0.0002269516,0.00005683472],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005283482,0.0004475529,0.006253892,0.0002651403,0.0002948382,0.0004386425,0.0003061517,0.3467905,0.02133322,0.01567633,0.02804541,0.57962],"study_design_scores_gemma":[0.00001521145,0.00006631015,0.0004588267,0.00001314852,0.00002281704,0.00007952612,0.00002475262,0.9878301,0.003819188,0.005996345,0.001656641,0.00001705806],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02614925,0.0005132188,0.9615144,0.0003691906,0.0001260338,0.00009908162,0.001132464,0.008562474,0.001533924],"genre_scores_gemma":[0.6147701,0.0006717255,0.3666146,0.0006954083,0.0001108384,0.0003691967,0.008184646,0.0009569906,0.007626525],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007285203,"threshold_uncertainty_score":0.0144856,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0359103846591871,"score_gpt":0.378476466764079,"score_spread":0.3425660821048919,"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."}}