{"id":"W4410872475","doi":"10.1101/2025.05.28.25328511","title":"PheCode-guided multi-modal topic modeling of electronic health records improves disease incidence prediction and GWAS discovery from UK Biobank","year":2025,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; McGill University","funders":"Fonds de recherche du Québec – Nature et technologies; Alliance de recherche numérique du Canada; Canada First Research Excellence Fund; Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Biobank; Health records; Modal; Data science; Disease; Incidence (geometry); Genome-wide association study; Data discovery; Medicine; Computer science; Internal medicine; Bioinformatics; Political science; World Wide Web; Health care; Biology; Mathematics; Genetics","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003137222,0.0002668235,0.0003978676,0.00006999724,0.00007223761,0.00003207626,0.0003142788,0.0003662297,0.000003924779],"category_scores_gemma":[0.0003812017,0.0002408893,0.0001335493,0.00006186694,0.0001537937,0.000004807038,0.000664698,0.0003593643,3.515148e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005249647,"about_ca_system_score_gemma":0.0008222445,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001267205,"about_ca_topic_score_gemma":0.0003944977,"domain_scores_codex":[0.9981257,0.0001219011,0.0004966374,0.0007520485,0.0001664798,0.0003372208],"domain_scores_gemma":[0.9989657,0.00003022031,0.0002275929,0.000568896,0.00006960306,0.0001379791],"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":[0.001449359,0.0008307487,0.6990126,0.003810967,0.00145508,0.00002068015,0.001064693,0.009542773,0.147087,0.0002408405,0.002126734,0.1333585],"study_design_scores_gemma":[0.004387165,0.001756883,0.3299154,0.003237227,0.0006029241,0.000009868098,0.0004591696,0.6125061,0.02922646,0.01187698,0.003927324,0.002094539],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9294161,0.01488526,0.05386715,0.0004474418,0.0005926955,0.0002562711,0.0004818155,0.00003259468,0.0000206854],"genre_scores_gemma":[0.9904124,0.005362588,0.002860174,0.0001756105,0.0002324658,0.00005096889,0.0005218523,0.00001380807,0.0003701905],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6029633,"threshold_uncertainty_score":0.9823182,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02675848719481826,"score_gpt":0.3081720775947838,"score_spread":0.2814135903999655,"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."}}