Simon Høj, Vibeke Backer, Charlotte Suppli Ulrik, Torben Sigsgaard, Howraman Meteran
{"title":"Artificial intelligence in asthma health literacy: a comparative analysis of ChatGPT versus Gemini.","authors":"Simon Høj, Vibeke Backer, Charlotte Suppli Ulrik, Torben Sigsgaard, Howraman Meteran","doi":"10.1080/02770903.2025.2495729","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Asthma is a complex and heterogeneous chronic disease affecting over 300 million individuals worldwide. Despite advances in pharmacotherapy, poor disease control remains a major challenge, necessitating innovative approaches to patient education and self-management. Artificial intelligence driven chatbots, such as ChatGPT and Gemini, have the potential to enhance asthma care by providing real-time, evidence-based information. As asthma management moves toward personalized medicine, AI could support individualized education and treatment guidance. However, concerns remain regarding the accuracy and reliability of AI-generated medical content.</p><p><strong>Objective: </strong>This study evaluated the accuracy of ChatGPT (version 4.0) and Gemini (version 1.2) in providing asthma-related health information using the Patient-completed Asthma Knowledge Questionnaire, a validated asthma literacy tool.</p><p><strong>Methods: </strong>A cross-sectional study was conducted in which both AI models answered 54 standardized asthma-related items. Responses were classified as correct or incorrect based on alignment with validated clinical knowledge. Accuracy was assessed using descriptive statistics, Cohen's kappa for inter-model agreement, and chi-square tests for comparative performance.</p><p><strong>Results: </strong>ChatGPT achieved an accuracy of 96.3% (52/54 correct; 95% CI: 87.5%-99.0%), while Gemini scored 92.6% (50/54 correct; 95% CI: 82.5%-97.1%), with no statistically significant difference (<i>p</i> = 0.67). Cohen's kappa demonstrated near-perfect agreement for ChatGPT (κ = 0.91) and strong agreement for Gemini (κ = 0.82).</p><p><strong>Conclusion: </strong>ChatGPT and Gemini demonstrated high accuracy in delivering asthma-related health information, supporting their potential as adjunct tools for patient education. AI models could potentially play a role in personalized asthma management by providing tailored treatment guidance and improving patient engagement.</p>","PeriodicalId":15076,"journal":{"name":"Journal of Asthma","volume":" ","pages":"1560-1566"},"PeriodicalIF":1.3000,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Asthma","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1080/02770903.2025.2495729","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/4/26 0:00:00","PubModel":"Epub","JCR":"Q3","JCRName":"ALLERGY","Score":null,"Total":0}
引用次数: 0
Abstract
Background: Asthma is a complex and heterogeneous chronic disease affecting over 300 million individuals worldwide. Despite advances in pharmacotherapy, poor disease control remains a major challenge, necessitating innovative approaches to patient education and self-management. Artificial intelligence driven chatbots, such as ChatGPT and Gemini, have the potential to enhance asthma care by providing real-time, evidence-based information. As asthma management moves toward personalized medicine, AI could support individualized education and treatment guidance. However, concerns remain regarding the accuracy and reliability of AI-generated medical content.
Objective: This study evaluated the accuracy of ChatGPT (version 4.0) and Gemini (version 1.2) in providing asthma-related health information using the Patient-completed Asthma Knowledge Questionnaire, a validated asthma literacy tool.
Methods: A cross-sectional study was conducted in which both AI models answered 54 standardized asthma-related items. Responses were classified as correct or incorrect based on alignment with validated clinical knowledge. Accuracy was assessed using descriptive statistics, Cohen's kappa for inter-model agreement, and chi-square tests for comparative performance.
Results: ChatGPT achieved an accuracy of 96.3% (52/54 correct; 95% CI: 87.5%-99.0%), while Gemini scored 92.6% (50/54 correct; 95% CI: 82.5%-97.1%), with no statistically significant difference (p = 0.67). Cohen's kappa demonstrated near-perfect agreement for ChatGPT (κ = 0.91) and strong agreement for Gemini (κ = 0.82).
Conclusion: ChatGPT and Gemini demonstrated high accuracy in delivering asthma-related health information, supporting their potential as adjunct tools for patient education. AI models could potentially play a role in personalized asthma management by providing tailored treatment guidance and improving patient engagement.
期刊介绍:
Providing an authoritative open forum on asthma and related conditions, Journal of Asthma publishes clinical research around such topics as asthma management, critical and long-term care, preventative measures, environmental counselling, and patient education.