A New Era in Medical Information: ChatGPT Outperforms Medical Information Provided by Online Information Sheets About Congenital Malformations.

Background: ChatGPT has demonstrated notable capabilities and has gained popularity in various medical tasks, including patient education. This study evaluates the content and readability of ChatGPT's responses to parents' questions about congenital anomalies.
Methods: Information on four congenital anomalies (congenital diaphragmatic hernia, esophageal atresia and tracheoesophageal fistula, anorectal malformation, and gastroschisis) was assessed. Seven questions frequently asked by parents were posed for each anomaly, and responses generated by GPT-4 were compared to online information sheets from three top pediatric medical centers. Two senior pediatric surgeons, blinded to the source, evaluated the answers based on accuracy, comprehensiveness, and conciseness. Reading time and readability of the answers were also assessed.
Results: ChatGPT answered all 28 questions, while online information sheets varied in completeness. ChatGPT's responses were rated significantly higher regarding full accuracy, comprehensiveness, and conciseness compared to the online information sheets (p < 0.00001, <0.00001, 0.0002, respectively). Despite having longer reading times and more challenging to read, ChatGPT's responses were more precise and detailed.
Conclusions: ChatGPT outperforms online information sheets in providing accurate, comprehensive, and concise answers about congenital anomalies. This positions ChatGPT as a beneficial supplementary resource in pediatric healthcare. Future research should explore real-world applications and usability among parents.
Level of Evidence: Level III.
Competing Interests: Conflict of interest The authors have no relevant financial or non-financial interests to disclose.
(Copyright © 2024 Elsevier Inc. All rights reserved.)