{"id":"W4385848332","doi":"10.1016/j.patter.2023.100887","title":"Enhancing phenotype recognition in clinical notes using large language models: PhenoBCBERT and PhenoGPT","year":2023,"lang":"en","type":"article","venue":"Patterns","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":65,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"U.S. National Library of Medicine; National Human Genome Research Institute; Intellectual and Developmental Disabilities Research Center; CHEO Research Institute; University of Pennsylvania; National Institutes of Health; Eunice Kennedy Shriver National Institute of Child Health and Human Development; Children's Hospital of Philadelphia","keywords":"Phenotype; Leverage (statistics); Computer science; Vocabulary; Heuristic; Ontology; Natural language processing; Scope (computer science); Artificial intelligence; Computational biology; Machine learning; Biology; Gene; Genetics; Linguistics","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004588095,0.00009971342,0.000144209,0.00005148096,0.00004190723,0.00001616325,0.00007523031,0.0001676571,0.00001676208],"category_scores_gemma":[0.0002882069,0.00009304118,0.00004080998,0.00009382785,0.00004382175,0.000004266009,0.0001285234,0.0001145303,0.00001748575],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000006562464,"about_ca_system_score_gemma":0.00002064781,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000133334,"about_ca_topic_score_gemma":0.0005281342,"domain_scores_codex":[0.9990546,0.00008157868,0.00023208,0.0002955235,0.00007676821,0.0002594976],"domain_scores_gemma":[0.9996604,0.0000686499,0.00004665547,0.0001462549,0.00001991513,0.00005814676],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0001946915,0.0002930982,0.2508776,0.0002420568,0.0001169976,0.0001049321,0.003510582,0.00009570776,0.2239174,0.00002763438,0.0003654237,0.520254],"study_design_scores_gemma":[0.01577066,0.002482406,0.5143098,0.00192041,0.0003104573,0.0001167197,0.02364029,0.1693165,0.2520135,0.007359368,0.008420099,0.004339773],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9865572,0.0004170697,0.01254277,0.00009173038,0.0001499532,0.00006562309,0.00004202293,0.00003858964,0.0000950606],"genre_scores_gemma":[0.997699,0.0002621569,0.001280639,0.0002839803,0.0002472136,0.000006762801,0.0001561724,0.00001621071,0.00004785987],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5159142,"threshold_uncertainty_score":0.3794109,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07803578452570843,"score_gpt":0.3586311829778933,"score_spread":0.2805953984521849,"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."}}