{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,8]],"date-time":"2026-05-08T07:01:45Z","timestamp":1778223705145,"version":"3.51.4"},"reference-count":15,"publisher":"Oxford University Press (OUP)","issue":"9","license":[{"start":{"date-parts":[[2024,5,21]],"date-time":"2024-05-21T00:00:00Z","timestamp":1716249600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/pages\/standard-publication-reuse-rights"}],"funder":[{"DOI":"10.13039\/100000050","name":"National Heart Lung and Blood Institute","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100000050","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100000002","name":"NIH","doi-asserted-by":"publisher","award":["5R01HL141841-05"],"award-info":[{"award-number":["5R01HL141841-05"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2024,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:sec>\n                  <jats:title>Background<\/jats:title>\n                  <jats:p>Artificial intelligence (AI) and large language models (LLMs) can play a critical role in emergency room operations by augmenting decision-making about patient admission. However, there are no studies for LLMs using real-world data and scenarios, in comparison to and being informed by traditional supervised machine learning (ML) models. We evaluated the performance of GPT-4 for predicting patient admissions from emergency department (ED) visits. We compared performance to traditional ML models both naively and when informed by few-shot examples and\/or numerical probabilities.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Methods<\/jats:title>\n                  <jats:p>We conducted a retrospective study using electronic health records across 7 NYC hospitals. We trained Bio-Clinical-BERT and XGBoost (XGB) models on unstructured and structured data, respectively, and created an ensemble model reflecting ML performance. We then assessed GPT-4 capabilities in many scenarios: through Zero-shot, Few-shot with and without retrieval-augmented generation (RAG), and with and without ML numerical probabilities.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>The Ensemble ML model achieved an area under the receiver operating characteristic curve (AUC) of 0.88, an area under the precision-recall curve (AUPRC) of 0.72 and an accuracy of 82.9%. The na\u00efve GPT-4's performance (0.79 AUC, 0.48 AUPRC, and 77.5% accuracy) showed substantial improvement when given limited, relevant data to learn from (ie, RAG) and underlying ML probabilities (0.87 AUC, 0.71 AUPRC, and 83.1% accuracy). Interestingly, RAG alone boosted performance to near peak levels (0.82 AUC, 0.56 AUPRC, and 81.3% accuracy).<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Conclusions<\/jats:title>\n                  <jats:p>The na\u00efve LLM had limited performance but showed significant improvement in predicting ED admissions when supplemented with real-world examples to learn from, particularly through RAG, and\/or numerical probabilities from traditional ML models. Its peak performance, although slightly lower than the pure ML model, is noteworthy given its potential for providing reasoning behind predictions. Further refinement of LLMs with real-world data is necessary for successful integration as decision-support tools in care settings.<\/jats:p>\n               <\/jats:sec>","DOI":"10.1093\/jamia\/ocae103","type":"journal-article","created":{"date-parts":[[2024,5,21]],"date-time":"2024-05-21T13:55:14Z","timestamp":1716299714000},"page":"1921-1928","source":"Crossref","is-referenced-by-count":73,"title":["Evaluating the accuracy of a state-of-the-art large language model for prediction of admissions from the emergency room"],"prefix":"10.1093","volume":"31","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4515-8090","authenticated-orcid":false,"given":"Benjamin S","family":"Glicksberg","sequence":"first","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Prem","family":"Timsina","sequence":"additional","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"Institute for Healthcare Delivery Science, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Dhaval","family":"Patel","sequence":"additional","affiliation":[{"name":"Institute for Healthcare Delivery Science, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1525-8541","authenticated-orcid":false,"given":"Ashwin","family":"Sawant","sequence":"additional","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Akhil","family":"Vaid","sequence":"additional","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Ganesh","family":"Raut","sequence":"additional","affiliation":[{"name":"Institute for Healthcare Delivery Science, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Alexander W","family":"Charney","sequence":"additional","affiliation":[{"name":"The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6217-1620","authenticated-orcid":false,"given":"Donald","family":"Apakama","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Icahn School of Medicine at Mount Sinai , New York, NY 10029, United States"}]},{"given":"Brendan G","family":"Carr","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Icahn School of Medicine at Mount Sinai , New York, NY 10029, United States"},{"name":"Department of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4946-6533","authenticated-orcid":false,"given":"Robert","family":"Freeman","sequence":"additional","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"Institute for Healthcare Delivery Science, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Girish N","family":"Nadkarni","sequence":"additional","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]},{"given":"Eyal","family":"Klang","sequence":"additional","affiliation":[{"name":"Division of Data-Driven and Digital Medicine, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"},{"name":"The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States"}]}],"member":"286","published-online":{"date-parts":[[2024,5,21]]},"reference":[{"issue":"1","key":"2024082207513375100_ocae103-B1","doi-asserted-by":"crossref","first-page":"669","DOI":"10.1186\/s12913-022-08070-7","article-title":"Influence of artificial intelligence on the work design of emergency department clinicians: a systematic literature review","volume":"22","author":"Boonstra","year":"2022","journal-title":"BMC Health Serv Res"},{"issue":"5","key":"2024082207513375100_ocae103-B2","doi-asserted-by":"crossref","first-page":"358","DOI":"10.1136\/emj.2010.103531","article-title":"Predicting emergency department admissions","volume":"29","author":"Boyle","year":"2012","journal-title":"Emerg Med J"},{"issue":"2","key":"2024082207513375100_ocae103-B3","doi-asserted-by":"crossref","first-page":"799","DOI":"10.1111\/poms.13580","article-title":"To predict or not to predict: the case of the emergency department","volume":"31","author":"Somanchi","year":"2022","journal-title":"Prod Oper Manag"},{"issue":"3","key":"2024082207513375100_ocae103-B4","doi-asserted-by":"crossref","first-page":"191","DOI":"10.1136\/emermed-2020-210902","article-title":"Predicting inhospital admission at the emergency department: a systematic review","volume":"39","author":"Brink","year":"2022","journal-title":"Emerg Med J"},{"key":"2024082207513375100_ocae103-B5","doi-asserted-by":"crossref","first-page":"101109","DOI":"10.1016\/j.ienj.2021.101109","article-title":"Machine learning methods applied to triage in emergency services: A systematic review","volume":"60","author":"S\u00e1nchez-Salmer\u00f3n","year":"2022","journal-title":"Int Emerg Nurs"},{"issue":"8","key":"2024082207513375100_ocae103-B6","doi-asserted-by":"crossref","first-page":"1930","DOI":"10.1038\/s41591-023-02448-8","article-title":"Large language models in medicine","volume":"29","author":"Thirunavukarasu","year":"2023","journal-title":"Nat Med"},{"issue":"1","key":"2024082207513375100_ocae103-B7","doi-asserted-by":"crossref","first-page":"194","DOI":"10.1038\/s41746-022-00742-2","article-title":"A large language model for electronic health records","volume":"5","author":"Yang","year":"2022","journal-title":"NPJ Digit Med"},{"issue":"7969","key":"2024082207513375100_ocae103-B8","doi-asserted-by":"crossref","first-page":"357","DOI":"10.1038\/s41586-023-06160-y","article-title":"Health system-scale language models are all-purpose prediction engines","volume":"619","author":"Jiang","year":"2023","journal-title":"Nature"},{"issue":"1","key":"2024082207513375100_ocae103-B9","doi-asserted-by":"crossref","first-page":"210","DOI":"10.1038\/s41746-023-00958-w","article-title":"A study of generative large language model for medical research and healthcare","volume":"6","author":"Peng","year":"2023","journal-title":"NPJ Digit Med"},{"key":"2024082207513375100_ocae103-B10","first-page":"9459","article-title":"Retrieval-augmented generation for knowledge-intensive NLP tasks","volume":"33","author":"Lewis","year":"2020","journal-title":"Adv Neural Inf Process Syst"},{"key":"2024082207513375100_ocae103-B11","author":"Gao","year":"2024"},{"key":"2024082207513375100_ocae103-B12","author":"Vaid","year":"2024"},{"key":"2024082207513375100_ocae103-B13","author":"Eriksen"},{"issue":"14","key":"2024082207513375100_ocae103-B14","doi-asserted-by":"crossref","first-page":"3717","DOI":"10.3390\/cancers15143717","article-title":"Enhancing triage efficiency and accuracy in emergency rooms for patients with metastatic prostate cancer: a retrospective analysis of artificial intelligence-assisted triage using ChatGPT 4.0","volume":"15","author":"Gebrael","year":"2023","journal-title":"Cancers"},{"key":"2024082207513375100_ocae103-B15","author":"Ling","year":"2023"}],"container-title":["Journal of the American Medical Informatics Association"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/jamia\/article-pdf\/31\/9\/1921\/58868088\/ocae103.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jamia\/article-pdf\/31\/9\/1921\/58868088\/ocae103.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,8,22]],"date-time":"2024-08-22T11:19:48Z","timestamp":1724325588000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jamia\/article\/31\/9\/1921\/7676138"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,5,21]]},"references-count":15,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2024,5,21]]},"published-print":{"date-parts":[[2024,9,1]]}},"URL":"https:\/\/doi.org\/10.1093\/jamia\/ocae103","relation":{},"ISSN":["1067-5027","1527-974X"],"issn-type":[{"value":"1067-5027","type":"print"},{"value":"1527-974X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2024,9]]},"published":{"date-parts":[[2024,5,21]]}}}