Diversifying Kidney Transplant Education: Assessing the Artificial Intelligence-Powered Capability of ChatGPT

Abstract

Background: Artificial intelligence (AI) has rapidly advanced, significantly impacting medicine. ChatGPT, a new AI model, generates responses based on user input. This study evaluates ChatGPT’s ability to assist with pre- and post-kidney transplantation (KT) patient education. Methods: ChatGPT was queried about KT on 21 February 2023 and 2 March 2023. Questions were categorized into general information for pre-KT patients or donors and post-KT patient instructions. Two experts independently assessed the accuracy of ChatGPT’s responses, and the Flesch–Kincaid readability test was applied to evaluate readability. Results: ChatGPT’s responses to general pre-KT questions were clear, concise, and accurate but occasionally misleading. Post-transplant instructions were generally clear and partially concise but lacked supporting evidence. Instructions for emergency situations post-KT were typically safe and reliable, whereas medication-related directions were often inaccurate and unreliable. The mean Flesch–Kincaid readability score was 30, indicating that ChatGPT’s answers were not easy to understand. Conclusion: This study demonstrates that while ChatGPT can provide clear definitions, explain symptoms, and offer reasonable advice on managing medical situations after KT, it frequently gives misleading answers to scientific inquiries. Transplantation researchers and providers should recognize ChatGPT as a potential information source for patients but exercise caution due to its incomplete accuracy and lack of references.