{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,2]],"date-time":"2026-06-02T14:09:52Z","timestamp":1780409392860,"version":"3.54.1"},"reference-count":30,"publisher":"Ovid Technologies (Wolters Kluwer Health)","issue":"7","content-domain":{"domain":["www.ahajournals.org"],"crossmark-restriction":true},"short-container-title":["JAHA"],"published-print":{"date-parts":[[2025,4]]},"abstract":"<jats:sec xml:lang=\"en\">\n                    <jats:title>Background<\/jats:title>\n                    <jats:p xml:lang=\"en\">Rates of oral anticoagulation (OAC) nonprescription in atrial fibrillation approach\u00a050%. Understanding reasons for OAC nonprescription may reduce gaps in guideline\u2010recommended care. We aimed to identify reasons for OAC nonprescription from clinical notes using large language models.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec xml:lang=\"en\">\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p xml:lang=\"en\">We identified all patients and associated clinical notes in our health care system with a clinician\u2010billed visit for atrial fibrillation without another indication for OAC and stratified them on the basis of active OAC prescriptions. Three annotators labeled reasons for OAC nonprescription in clinical notes on 10% of all patients (\u201cannotation set\u201d). We engineered prompts for a generative large language model (Generative Pre\u2010trained Transformer 4) and trained a discriminative large language model (ClinicalBERT) to identify reasons for OAC nonprescription and selected the best\u2010performing model to predict reasons for the remaining 90% of patients (\u201cinference set\u201d).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec xml:lang=\"en\">\n                    <jats:title>Results<\/jats:title>\n                    <jats:p xml:lang=\"en\">A total of 35\u2009737 patients were identified, of which 7712 (21.6%) did not have active OAC prescriptions. A total of 910 notes across 771 patients were annotated. Generative Pre\u2010trained Transformer 4 outperformed ClinicalBERT (macro\u2010F1 score across all reasons of 0.79, compared with 0.69 for ClinicalBERT). Using Generative Pre\u2010trained Transformer 4 on the inference set, 61.1% of notes had documented reasons for OAC nonprescription, most commonly the alternative use of an antiplatelet agent (23.3%), therapeutic inertia (21.0%), and low burden of atrial fibrillation (17.1%).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec xml:lang=\"en\">\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p xml:lang=\"en\">This is the first study using large language models to extract documented reasons for OAC nonprescription from clinical notes in patients with atrial fibrillation and reveals guideline\u2010discordant practices and actionable insights for the development of health system interventions to reduce OAC nonprescription.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1161\/jaha.124.040419","type":"journal-article","created":{"date-parts":[[2025,3,27]],"date-time":"2025-03-27T06:11:19Z","timestamp":1743055879000},"update-policy":"https:\/\/doi.org\/10.1161\/crossmarkpolicy","source":"Crossref","is-referenced-by-count":4,"title":["Understanding Reasons for Oral Anticoagulation Nonprescription in Atrial Fibrillation Using Large Language Models"],"prefix":"10.1161","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0913-8674","authenticated-orcid":false,"given":"Sulaiman","family":"Somani","sequence":"first","affiliation":[{"name":"Stanford University","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1987-3680","authenticated-orcid":false,"given":"Dale Daniel","family":"Kim","sequence":"additional","affiliation":[{"name":"Stanford University","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-1899-7074","authenticated-orcid":false,"given":"Eduardo","family":"Perez\u2010Guerrero","sequence":"additional","affiliation":[{"name":"Stanford University","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7434-4143","authenticated-orcid":false,"given":"Summer","family":"Ngo","sequence":"additional","affiliation":[{"name":"Stanford University","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7937-9564","authenticated-orcid":false,"given":"Tina","family":"Seto","sequence":"additional","affiliation":[{"name":"Stanford Health Care","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1122-7695","authenticated-orcid":false,"given":"Sadeer","family":"Al\u2010Kindi","sequence":"additional","affiliation":[{"name":"Houston Methodist","place":["Houston, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6553-3455","authenticated-orcid":false,"given":"Tina","family":"Hernandez\u2010Boussard","sequence":"additional","affiliation":[{"name":"Stanford University","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5226-0723","authenticated-orcid":false,"given":"Fatima","family":"Rodriguez","sequence":"additional","affiliation":[{"name":"Stanford University","place":["Stanford, USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"276","reference":[{"key":"e_1_3_2_2_2","doi-asserted-by":"publisher","DOI":"10.1161\/CIR.0000000000001193"},{"key":"e_1_3_2_3_2","doi-asserted-by":"publisher","DOI":"10.1161\/CIR.0000000000001209"},{"key":"e_1_3_2_4_2","doi-asserted-by":"publisher","DOI":"10.1093\/eurheartj\/ehz386"},{"key":"e_1_3_2_5_2","doi-asserted-by":"publisher","DOI":"10.1001\/jamacardio.2015.0374"},{"key":"e_1_3_2_6_2","doi-asserted-by":"publisher","DOI":"10.1161\/JAHA.122.028120"},{"key":"e_1_3_2_7_2","doi-asserted-by":"publisher","DOI":"10.1038\/s43856-022-00157-w"},{"key":"e_1_3_2_8_2","doi-asserted-by":"publisher","DOI":"10.1378\/chest.09-1584"},{"key":"e_1_3_2_9_2","unstructured":"Xiao S Liu Z Zhang P Muennighof N. C\u2010Pack: packaged resources to advance general Chinese embedding. arXiv [cs.CL]. 2023 http:\/\/arxiv.org\/abs\/2309.07597."},{"key":"e_1_3_2_10_2","doi-asserted-by":"publisher","DOI":"10.21203\/rs.3.rs-5095287\/v1"},{"key":"e_1_3_2_11_2","doi-asserted-by":"publisher","DOI":"10.1038\/sdata.2016.35"},{"key":"e_1_3_2_12_2","unstructured":"Szyma\u0144ski P Kajdanowicz T. A network perspective on stratification of multi\u2010label data. arXiv [StatML]2017:22\u201335."},{"key":"e_1_3_2_13_2","doi-asserted-by":"publisher","DOI":"10.1093\/jamia\/ocae146"},{"key":"e_1_3_2_14_2","doi-asserted-by":"publisher","DOI":"10.7326\/0003-4819-131-7-199910050-00003"},{"key":"e_1_3_2_15_2","doi-asserted-by":"publisher","DOI":"10.1093\/ehjopen\/oead051"},{"key":"e_1_3_2_16_2","doi-asserted-by":"publisher","DOI":"10.1161\/CIRCULATIONAHA.118.035909"},{"key":"e_1_3_2_17_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.ahj.2018.03.003"},{"key":"e_1_3_2_18_2","doi-asserted-by":"publisher","DOI":"10.1161\/jaha.121.024868"},{"key":"e_1_3_2_19_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.jacc.2016.11.061"},{"key":"e_1_3_2_20_2","doi-asserted-by":"publisher","DOI":"10.1177\/08971900211004840"},{"key":"e_1_3_2_21_2","doi-asserted-by":"publisher","DOI":"10.1093\/jamia\/ocz119"},{"key":"e_1_3_2_22_2","doi-asserted-by":"publisher","DOI":"10.1001\/jamanetworkopen.2023.9638"},{"key":"e_1_3_2_23_2","doi-asserted-by":"publisher","DOI":"10.1161\/CIR.0000000000000040"},{"key":"e_1_3_2_24_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.jacc.2006.07.009"},{"key":"e_1_3_2_25_2","doi-asserted-by":"publisher","DOI":"10.1161\/CIR.0000000000000568"},{"key":"e_1_3_2_26_2","doi-asserted-by":"publisher","DOI":"10.12998\/wjcc.v10.i3.939"},{"key":"e_1_3_2_27_2","doi-asserted-by":"publisher","DOI":"10.1001\/jamacardio.2018.1176"},{"key":"e_1_3_2_28_2","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMoa2310234"},{"key":"e_1_3_2_29_2","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMoa2303062"},{"key":"e_1_3_2_30_2","unstructured":"Epic and Microsoft bring GPT\u20104 to EHRs. Accessed October 8 2024. https:\/\/www.epic.com\/epic\/post\/epic\u2010and\u2010microsoft\u2010bring\u2010gpt\u20104\u2010to\u2010ehrs\/."},{"key":"e_1_3_2_31_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-023-02552-9"}],"container-title":["Journal of the American Heart Association"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.ahajournals.org\/doi\/full\/10.1161\/JAHA.124.040419","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,11,24]],"date-time":"2025-11-24T13:38:10Z","timestamp":1763991490000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ahajournals.org\/doi\/10.1161\/JAHA.124.040419"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,4]]},"references-count":30,"journal-issue":{"issue":"7","published-print":{"date-parts":[[2025,4]]}},"alternative-id":["10.1161\/JAHA.124.040419"],"URL":"https:\/\/doi.org\/10.1161\/jaha.124.040419","relation":{},"ISSN":["2047-9980"],"issn-type":[{"value":"2047-9980","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,4]]},"assertion":[{"value":"2024-12-02","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-01-28","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-03-27","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e040419"}}