Artificial intelligence performance in pediatric asthma

Asthma is the most common chronic disease of childhood, characterized by symptoms such as wheezing, shortness of breath, and coughing. With the advancement of technology, artificial intelligence (AI) applications are increasingly being used in various fields, among which ChatGPT is one of the most widely utilized. The aim of this study is to evaluate the reliability, quality, and readability of the answers provided by the ChatGPT-4o application to questions related to pediatric asthma.

The ChatGPT-4o application was used to record answers to 25 of the most frequently asked questions about asthma in children. To determine the quality and reliability of the answers, we used the Global Quality Scale and modified DISCERN tool. We tested readability using seven indices: Automated Readability Index, Flesch Reading Ease Score, Flesch-Kincaid Grade Level (FKGL), Gunning Fog Readability Index, Simple Measure of Gobbledygook, Coleman-Liau Readability Index, and Linsear Write Formula.

The answers provided by the ChatGPT-4o application to questions about childhood asthma were found to have good reliability (88% by the first evaluator and 84% by the second evaluator) and high quality (88% by both evaluators). The application scored 10.77 ± 1.58 on the FKGL scale, and in conjunction with the other indices, the results indicated that the answers required a high level of reading proficiency.

Artificial intelligence can be a reliable tool for parents in providing information about pediatric asthma. However, these findings suggest that readability issues may hinder the clinical application of AI-generated content in asthma diagnosis and treatment.