{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,17]],"date-time":"2026-11-17T11:58:41Z","timestamp":1794916721318,"version":"4.9.1"},"reference-count":20,"publisher":"Springer Science and Business Media LLC","issue":"5","license":[{"start":{"date-parts":[[2026,2,23]],"date-time":"2026-02-23T00:00:00Z","timestamp":1771804800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,3,12]],"date-time":"2026-03-12T00:00:00Z","timestamp":1773273600000},"content-version":"vor","delay-in-days":17,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Nat Med"],"published-print":{"date-parts":[[2026,5]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>ChatGPT Health was launched in January 2026 as OpenAI\u2019s consumer health tool and has reached millions of users. Here we conducted a structured stress test of triage recommendations using 60 clinician-authored vignettes across 21 clinical domains under 16 factorial conditions, yielding 960 total responses. Performance followed an inverted U-shaped pattern, with the most dangerous failures concentrated at clinical extremes\u2014nonurgent presentations (35%) and emergency conditions (48%). Among gold-standard emergencies, the system undertriaged 52% of cases, directing patients with diabetic ketoacidosis or impending respiratory failure to 24\u201348\u2009h evaluation rather than the emergency department, while correctly triaging classical emergencies such as stroke and anaphylaxis. When family or friends minimized symptoms, indicating anchoring bias, triage recommendations shifted significantly in edge cases (odds ratio\u2009=\u200911.7, 95% confidence interval\u2009=\u20093.7\u201336.6), with the majority of shifts toward less urgent care. Crisis-intervention messages activated unpredictably across suicidal ideation presentations, occurring more frequently when patients described no specific method than when they did. Patient race, sex and barriers to care did not show significant effects, although confidence intervals did not exclude clinically meaningful differences. These findings reveal missed high-risk emergencies and inconsistent activation of crisis safeguards, raising safety concerns that warrant prospective validation before consumer-scale deployment of artificial intelligence triage systems.<\/jats:p>","DOI":"10.1038\/s41591-026-04297-7","type":"journal-article","created":{"date-parts":[[2026,2,23]],"date-time":"2026-02-23T15:34:46Z","timestamp":1771860886000},"page":"1671-1675","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":101,"title":["ChatGPT Health performance in a structured test of triage recommendations"],"prefix":"10.1038","volume":"32","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-8816-7838","authenticated-orcid":false,"given":"Ashwin","family":"Ramaswamy","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alvira","family":"Tyagi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hannah","family":"Hugo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Joy","family":"Jiang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pushkala","family":"Jayaraman","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mateen","family":"Jangda","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alexis E.","family":"Te","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Steven A.","family":"Kaplan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0081-0356","authenticated-orcid":false,"given":"Joshua","family":"Lampert","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4946-6533","authenticated-orcid":false,"given":"Robert","family":"Freeman","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nicholas","family":"Gavin","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3146-4524","authenticated-orcid":false,"given":"Ashutosh K.","family":"Tewari","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2045-518X","authenticated-orcid":false,"given":"Ankit","family":"Sakhuja","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bilal","family":"Naved","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8135-6858","authenticated-orcid":false,"given":"Alexander W.","family":"Charney","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0001-0438-0827","authenticated-orcid":false,"given":"Mahmud","family":"Omar","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Michael A.","family":"Gorin","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4567-3108","authenticated-orcid":false,"given":"Eyal","family":"Klang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6319-4314","authenticated-orcid":false,"given":"Girish N.","family":"Nadkarni","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,2,23]]},"reference":[{"key":"4297_CR1","unstructured":"OpenAI. Introducing ChatGPT Health. https:\/\/openai.com\/index\/introducing-chatgpt-health\/ (accessed 13 January 2026)."},{"key":"4297_CR2","doi-asserted-by":"publisher","first-page":"e6","DOI":"10.1016\/j.landig.2024.12.002","volume":"7","author":"V Bellini","year":"2025","unstructured":"Bellini, V. & Bignami, E. G. Generative pre-trained transformer 4 (GPT-4) in clinical settings. Lancet Digit. Health 7, e6\u2013e7 (2025).","journal-title":"Lancet Digit. Health"},{"key":"4297_CR3","doi-asserted-by":"publisher","DOI":"10.1038\/s43856-024-00717-2","volume":"5","author":"F Busch","year":"2025","unstructured":"Busch, F. et al. Current applications and challenges in large language models for patient care: a systematic review. Commun. Med. (Lond.) 5, 26 (2025).","journal-title":"Commun. Med. (Lond.)"},{"key":"4297_CR4","doi-asserted-by":"publisher","DOI":"10.1056\/AIoa2300015","volume":"2","author":"S Shekar","year":"2025","unstructured":"Shekar, S., Pataranutaporn, P., Sarabu, C., Cecchi, G. A. & Maes, P. People overtrust AI-generated medical advice despite low accuracy. NEJM AI 2, AIoa2300015 (2025).","journal-title":"NEJM AI"},{"key":"4297_CR5","doi-asserted-by":"publisher","first-page":"e441","DOI":"10.1016\/S2589-7500(24)00111-0","volume":"6","author":"A De Hond","year":"2024","unstructured":"De Hond, A. et al. From text to treatment: the crucial role of validation for generative large language models in health care. Lancet Digit. Health 6, e441\u2013e443 (2024).","journal-title":"Lancet Digit. Health"},{"key":"4297_CR6","doi-asserted-by":"publisher","DOI":"10.1056\/AIpc2400927","volume":"2","author":"D Blumenthal","year":"2025","unstructured":"Blumenthal, D. & Goldberg, C. Managing patient use of generative health AI. NEJM AI 2, AIpc2400927 (2025).","journal-title":"NEJM AI"},{"key":"4297_CR7","doi-asserted-by":"publisher","first-page":"1873","DOI":"10.1038\/s41591-025-03626-6","volume":"31","author":"M Omar","year":"2025","unstructured":"Omar, M. et al. Sociodemographic biases in medical decision making by large language models. Nat. Med. 31, 1873\u20131881 (2025).","journal-title":"Nat. Med."},{"key":"4297_CR8","unstructured":"Arora, R. K. et al. HealthBench: evaluating large language models towards improved human health. Preprint at https:\/\/arxiv.org\/abs\/2505.08775 (2025)."},{"key":"4297_CR9","doi-asserted-by":"publisher","first-page":"2613","DOI":"10.1038\/s41591-024-03097-1","volume":"30","author":"P Hager","year":"2024","unstructured":"Hager, P. et al. Evaluation and mitigation of the limitations of large language models in clinical decision-making. Nat. Med. 30, 2613\u20132622 (2024).","journal-title":"Nat. Med."},{"key":"4297_CR10","doi-asserted-by":"publisher","first-page":"924","DOI":"10.1038\/s41591-022-01772-9","volume":"28","author":"B Vasey","year":"2022","unstructured":"Vasey, B. et al. Reporting guideline for the early-stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. Nat. Med. 28, 924\u2013933 (2022).","journal-title":"Nat. Med."},{"key":"4297_CR11","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01684-1","volume":"8","author":"F Gaber","year":"2025","unstructured":"Gaber, F. et al. Evaluating large language model workflows in clinical decision support for triage and referral and diagnosis. NPJ Digit. Med. 8, 263 (2025).","journal-title":"NPJ Digit. Med."},{"key":"4297_CR12","doi-asserted-by":"publisher","first-page":"790","DOI":"10.1007\/s11606-024-09177-9","volume":"40","author":"HG Schmidt","year":"2025","unstructured":"Schmidt, H. G., Rotgans, J. I. & Mamede, S. Bias sensitivity in diagnostic decision-making: comparing ChatGPT with residents. J. Gen. Intern. Med. 40, 790\u2013795 (2025).","journal-title":"J. Gen. Intern. Med."},{"key":"4297_CR13","unstructured":"Thapa, R. et al. Disentangling reasoning and knowledge in medical large language models. Preprint at https:\/\/arxiv.org\/abs\/2505.11462 (2025)."},{"key":"4297_CR14","doi-asserted-by":"publisher","DOI":"10.1056\/AIdbp2500120","volume":"2","author":"LG McCoy","year":"2025","unstructured":"McCoy, L. G. et al. Assessment of large language models in clinical reasoning: a novel benchmarking study. NEJM AI 2, AIdbp2500120 (2025).","journal-title":"NEJM AI"},{"key":"4297_CR15","doi-asserted-by":"publisher","first-page":"50","DOI":"10.1518\/hfes.46.1.50.30392","volume":"46","author":"JD Lee","year":"2004","unstructured":"Lee, J. D. & See, K. A. Trust in automation: designing for appropriate reliance. Hum. Factors 46, 50\u201380 (2004).","journal-title":"Hum. Factors"},{"key":"4297_CR16","unstructured":"OpenAI. Helping people when they need it most. https:\/\/openai.com\/index\/helping-people-when-they-need-it-most (accessed 31 January 2026)."},{"key":"4297_CR17","doi-asserted-by":"publisher","first-page":"609","DOI":"10.1038\/s41591-025-04074-y","volume":"32","author":"AM Bean","year":"2026","unstructured":"Bean, A. M. et al. Reliability of LLMs as medical assistants for the general public: a randomized preregistered study. Nat. Med. 32, 609\u2013615 (2026).","journal-title":"Nat. Med."},{"key":"4297_CR18","doi-asserted-by":"publisher","first-page":"264","DOI":"10.1093\/ije\/dyt221","volume":"43","author":"FI Gunasekara","year":"2014","unstructured":"Gunasekara, F. I., Richardson, K., Carter, K. & Blakely, T. Fixed effects analysis of repeated measures data. Int. J. Epidemiol. 43, 264\u2013269 (2014).","journal-title":"Int. J. Epidemiol."},{"key":"4297_CR19","doi-asserted-by":"publisher","first-page":"60","DOI":"10.1038\/s41591-024-03425-5","volume":"31","author":"J Gallifant","year":"2025","unstructured":"Gallifant, J. et al. The TRIPOD-LLM reporting guideline for studies using large language models. Nat. Med. 31, 60\u201369 (2025).","journal-title":"Nat. Med."},{"key":"4297_CR20","doi-asserted-by":"publisher","unstructured":"ashwinra-code. ashwinra-code\/gpt-health-eval: v1.0 submission (v1.0). Zenodo https:\/\/doi.org\/10.5281\/zenodo.18451491 (2026).","DOI":"10.5281\/zenodo.18451491"}],"container-title":["Nature Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41591-026-04297-7.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41591-026-04297-7","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41591-026-04297-7.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,20]],"date-time":"2026-05-20T14:03:45Z","timestamp":1779285825000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41591-026-04297-7"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,23]]},"references-count":20,"journal-issue":{"issue":"5","published-print":{"date-parts":[[2026,5]]}},"alternative-id":["4297"],"URL":"https:\/\/doi.org\/10.1038\/s41591-026-04297-7","relation":{},"ISSN":["1078-8956","1546-170X"],"issn-type":[{"value":"1078-8956","type":"print"},{"value":"1546-170X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,23]]},"assertion":[{"value":"15 January 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"20 February 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"23 February 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"2 April 2026","order":5,"name":"change_date","label":"Change Date","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"Update","order":6,"name":"change_type","label":"Change Type","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"This article was originally published under the subscription model but it is now published under an Open Access license.","order":7,"name":"change_details","label":"Change Details","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"The authors declare no competing interests.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}]}}