The efficacy of ChatGPT: Responding to atopic dermatitis patient inquiries in traditional Chinese using a generative artificial intelligence model

This is the authors' abstract. We don't add key points for this paper.

Background: Atopic dermatitis (AD) is a common chronic skin condition that affects various aspects of daily life and can significantly reduce the quality of life for some patients. Effective patient education is essential for improving treatment adherence and outcomes. Recently, tools like ChatGPT have emerged as a new resource for addressing patient inquiries about AD. Objectives: This study aims to evaluate whether ChatGPT can effectively incorporate the latest AD treatments and accurately respond to patient questions in traditional Chinese. Methods: Forty-seven common patient queries about AD were input into ChatGPT-4 in traditional Chinese. A panel of national dermatologists reviewed the responses, assessing each on a 5-point Likert scale. Detailed comments accompanied each rating to provide further insights. Results: The average scores for ChatGPT’s responses ranged from 3.25 to 5, with common critiques including the use of simplified Chinese instead of traditional Chinese and the omission of the latest treatments. Dermatologists also noted inaccuracies in advice that strayed from evidence-based medicine. Conclusion: While ChatGPT shows promise as an educational tool for AD patients, concerns about the quality of translation and incorporation of new treatments remain. Further research is needed to ensure ChatGPT can provide reliable, up-to-date medical advice in non-English languages.