{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,13]],"date-time":"2026-11-13T16:27:19Z","timestamp":1794587239717,"version":"4.9.1"},"reference-count":20,"publisher":"Ovid Technologies (Wolters Kluwer Health)","issue":"25","license":[{"start":{"date-parts":[[2026,6,19]],"date-time":"2026-06-19T00:00:00Z","timestamp":1781827200000},"content-version":"unspecified","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":["lww.com","ovid.com"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2026,6,19]]},"abstract":"<jats:sec>\n                    <jats:title\/>\n                    <jats:p>This study aims to compare emergency physicians\u2019 performance with that of general-purpose large language models (LLMs), such as ChatGPT and Gemini, for pneumothorax (PTX) detection on chest radiographs (CXRs). This single-center, retrospective study of adults was conducted between January 2015 and February 2025 and included 265 PTX cases and 267 non-PTX controls. Exclusions included diagnoses made only by computed tomography, absence of CXR, initial treatment at another center, or incomplete data. Thirteen emergency physicians independently and blindly reviewed CXRs and recorded a binary decision. ChatGPT and Gemini evaluated the same images with a standardized yes\/no prompt, with memory cleared between cases to prevent carryover. The primary outcome was LLM diagnostic performance for PTX, while the secondary outcome compared LLMs with physicians. ChatGPT and Gemini exhibited distinct diagnostic performance profiles for PTX detection on CXRs. Gemini demonstrated a sensitivity of 52.5%, whereas ChatGPT demonstrated a sensitivity of 44.5%. Conversely, ChatGPT achieved a specificity of 95.5% and an overall accuracy of 70.1%, while Gemini demonstrated a specificity of 79.0% and an accuracy of 65.8%. Agreement with the reference standard was moderate for ChatGPT, with a kappa value of 0.401, and fair for Gemini, with a kappa value of 0.315. Increasing case difficulty was associated with a reduction in diagnostic accuracy for both models, with correlation coefficients of\u2005\u2212\u20050.438 for ChatGPT and\u2005\u2212\u20050.274 for Gemini. For contextual clinical comparison, emergency physicians demonstrated a sensitivity of 64.5%, a specificity of 99.6%, and an overall accuracy of 82.1%. This study demonstrates model-specific differences in PTX detection by general-purpose AI systems, with Gemini showing higher sensitivity and ChatGPT showing superior specificity and accuracy, both declining with increasing case difficulty. Physician performance remained higher, but was secondary for context. Despite their accessibility and low cost, these models should be considered only adjunctive tools until task-specific optimization and clinical validation are achieved.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1097\/md.0000000000049412","type":"journal-article","created":{"date-parts":[[2026,6,22]],"date-time":"2026-06-22T06:01:20Z","timestamp":1782108080000},"page":"e49412","update-policy":"https:\/\/doi.org\/10.1097\/lww.0000000000001000","source":"Crossref","is-referenced-by-count":0,"title":["Comparison of emergency physicians and artificial intelligence models in pneumothorax detection: A multi-reader retrospective study"],"prefix":"10.1097","volume":"105","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8254-6821","authenticated-orcid":false,"given":"Omer Faruk","family":"Cakiroglu","sequence":"first","affiliation":[{"name":"Department of Emergency Medicine, Kartal Doktor Lutfi Kirdar City Hospital, Istanbul, Turkiye"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2595-931X","authenticated-orcid":false,"given":"Bilal","family":"Arac","sequence":"additional","affiliation":[{"name":"Basaksehir State Hospital, Istanbul, Turkiye"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-9904-0893","authenticated-orcid":false,"given":"Seyma Nur","family":"Calisir","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Basaksehir Cam Sakura City Hospital, Istanbul, Turkey"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6690-1934","authenticated-orcid":false,"given":"Julide","family":"Gurbuz","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Bezmialem Vakif University, Istanbul, Turkiye."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0469-0593","authenticated-orcid":false,"given":"Elif Betul","family":"Balci","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Bezmialem Vakif University, Istanbul, Turkiye."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5920-8127","authenticated-orcid":false,"given":"Bahadir","family":"Taslidere","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Bezmialem Vakif University, Istanbul, Turkiye."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1035-1907","authenticated-orcid":false,"given":"Basar","family":"Cander","sequence":"additional","affiliation":[{"name":"Department of Emergency Medicine, Bezmialem Vakif University, Istanbul, Turkiye."}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"276","reference":[{"issue":"Suppl 4","key":"R1-20261113","first-page":"S372","article-title":"Pneumothorax: from definition to diagnosis and treatment.","volume":"6","author":"Zarogoulidis","year":"2014","journal-title":"J Thorac Dis"},{"key":"R2-20261113","doi-asserted-by":"crossref","first-page":"313","DOI":"10.2214\/ajr.166.2.8553937","article-title":"Comparison of upright inspiratory and expiratory chest radiographs for detecting pneumothoraces.","volume":"166","author":"Seow","year":"1996","journal-title":"AJR Am J Roentgenol"},{"key":"R11-20261113","doi-asserted-by":"crossref","first-page":"e2022779","DOI":"10.1001\/jamanetworkopen.2020.22779","article-title":"Comparison of chest radiograph interpretations by artificial intelligence algorithm vs radiology residents.","volume":"3","author":"Wu","year":"2020","journal-title":"JAMA Netw Open"},{"key":"R12-20261113","doi-asserted-by":"crossref","first-page":"e220510","DOI":"10.1148\/radiol.220510","article-title":"Artificial intelligence for clinical interpretation of bedside chest radiographs.","volume":"307","author":"Khader","year":"2023","journal-title":"Radiology"},{"key":"R13-20261113","doi-asserted-by":"crossref","first-page":"626369","DOI":"10.3389\/fmed.2021.626369","article-title":"A deep learning based framework for diagnosing multiple skin diseases in a clinical environment.","volume":"8","author":"Zhu","year":"2021","journal-title":"Front Med (Lausanne)"},{"key":"R14-20261113","doi-asserted-by":"crossref","first-page":"315","DOI":"10.1038\/s41591-024-03301-2","article-title":"Clinical validation of an AI-based pathology tool for scoring of metabolic dysfunction-associated steatohepatitis.","volume":"31","author":"Pulaski","year":"2024","journal-title":"Nat Med"},{"key":"R15-20261113","doi-asserted-by":"crossref","first-page":"233341","DOI":"10.1148\/radiol.233341","article-title":"Interobserver agreement and performance of concurrent AI assistance for radiographic evaluation of knee osteoarthritis.","volume":"312","author":"Brejneb\u00f8l","year":"2024","journal-title":"Radiology"},{"key":"R16-20261113","doi-asserted-by":"crossref","first-page":"3462","DOI":"10.1109\/CVPR.2017.369","article-title":"ChestX-ray8: hospital-scale chest X-ray database and benchmarks on weakly supervised classification and localization of common thorax diseases.","volume":"2017","author":"Wang","year":"2017","journal-title":"2017 IEEE Conf Comp Vision Pattern Recognition (CVPR)"},{"key":"R17-20261113","doi-asserted-by":"crossref","first-page":"194","DOI":"10.3390\/jpm15050194","article-title":"The accuracy of ChatGPT-4o in interpreting chest and abdominal X-Ray images.","volume":"15","author":"Lacaita","year":"2025","journal-title":"J Pers Med"},{"key":"R18-20261113","doi-asserted-by":"crossref","first-page":"220259","DOI":"10.1183\/16000617.0259-2022","article-title":"Deep learning for pneumothorax diagnosis: a systematic review and meta-analysis.","volume":"32","author":"Sugibayashi","year":"2023","journal-title":"Eur Respir Rev"},{"key":"R19-20261113","doi-asserted-by":"crossref","first-page":"7888","DOI":"10.1007\/s00330-021-07833-w","article-title":"Pneumothorax detection in chest radiographs: optimizing artificial intelligence system for accuracy and confounding bias reduction using in-image annotations in algorithm training.","volume":"31","author":"Rueckel","year":"2021","journal-title":"Eur Radiol"},{"key":"R20-20261113","doi-asserted-by":"crossref","first-page":"196","DOI":"10.1148\/radiol.2018180921","article-title":"Automated triaging of adult chest radiographs with deep artificial neural networks.","volume":"291","author":"Annarumma","year":"2019","journal-title":"Radiology"},{"key":"R21-20261113","doi-asserted-by":"crossref","first-page":"4099","DOI":"10.2147\/JMDH.S535405","article-title":"Multidisciplinary evaluation of an AI-based pneumothorax detection model: clinical comparison with physicians in edge and cloud environments.","volume":"18","author":"Dal","year":"2025","journal-title":"J Multidiscip Healthc"},{"key":"R22-20261113","doi-asserted-by":"crossref","first-page":"99","DOI":"10.1016\/j.ajem.2025.03.060","article-title":"Evaluating a large language model\u2019s accuracy in chest X-ray interpretation for acute thoracic conditions.","volume":"93","author":"Ostrovsky","year":"2025","journal-title":"Am J Emerg Med"},{"key":"R23-20261113","doi-asserted-by":"crossref","first-page":"19624","DOI":"10.1038\/s41598-024-70780-1","article-title":"Factors for increasing positive predictive value of pneumothorax detection on chest radiographs using artificial intelligence.","volume":"14","author":"Lee","year":"2024","journal-title":"Sci Rep"},{"key":"R24-20261113","doi-asserted-by":"crossref","first-page":"2038","DOI":"10.1016\/j.ejrad.2015.07.003","article-title":"Pneumothorax size measurements on digital chest radiographs: intra- and inter-rater reliability.","volume":"84","author":"Thelle","year":"2015","journal-title":"Eur J Radiol"},{"key":"R25-20261113","doi-asserted-by":"crossref","first-page":"4374","DOI":"10.1007\/s00330-024-11339-6","article-title":"CXR-LLaVA: a multimodal large language model for interpreting chest X-ray images.","volume":"35","author":"Lee","year":"2025","journal-title":"Eur Radiol"},{"key":"R26-20261113","doi-asserted-by":"crossref","first-page":"e0331962","DOI":"10.1371\/journal.pone.0331962","article-title":"New frontiers in radiologic interpretation: evaluating the effectiveness of large language models in pneumothorax diagnosis.","volume":"20","author":"Bulut","year":"2025","journal-title":"PLoS One"},{"key":"R27-20261113","doi-asserted-by":"crossref","first-page":"376","DOI":"10.3390\/diagnostics16030376","article-title":"Limitations in chest X-ray interpretation by vision-capable large language models, Gemini 1.0, Gemini 1.5 Pro, GPT-4 turbo, and GPT-4o.","volume":"16","author":"Chen","year":"2026","journal-title":"Diagnostics (Basel)"},{"key":"R28-20261113","doi-asserted-by":"crossref","first-page":"100664","DOI":"10.1016\/j.ejro.2025.100664","article-title":"Development of an AI model for pneumothorax imaging: dataset and model optimization strategies for real-world deployment.","volume":"14","author":"Tseng","year":"2025","journal-title":"Eur J Radiol Open"}],"container-title":["Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.ovid.com\/10.1097\/MD.0000000000049412","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,11,13]],"date-time":"2026-11-13T16:01:57Z","timestamp":1794585717000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ovid.com\/10.1097\/MD.0000000000049412"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,19]]},"references-count":20,"journal-issue":{"issue":"25","published-print":{"date-parts":[[2026]]}},"URL":"https:\/\/doi.org\/10.1097\/md.0000000000049412","relation":{},"ISSN":["0025-7974","1536-5964"],"issn-type":[{"value":"0025-7974","type":"print"},{"value":"1536-5964","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,19]]},"assertion":[{"value":"2026-02-12","name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-06-05","name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}}]}}