Readability of Professional Medical Content on Acute Appendicitis: A Comparative Cross-Sectional Study of ChatGPT and UpToDate

Abstract

Introduction Acute appendicitis is one of the most common diseases occurring due to inflammation of the vermiform appendix, which requires surgical intervention. With the advent of standardized artificial intelligence (AI) tools such as ChatGPT (OpenAI, San Francisco, CA), AI-based search engines have emerged as a secondary means for patients to educate themselves about their health. The readability of each response is important for concept understanding as well as the impact of novel therapeutics. Aims This study aims to evaluate and compare the readability of medical information on acute appendicitis generated by an AI language model and UpToDate (Wolters Kluwer Health, Waltham, MA) using established readability metrics. Methodology A comparative cross-sectional study was conducted to evaluate the readability of six ChatGPT-4o and six UpToDate responses on acute appendicitis. Readability parameters were assessed using WebFX (WebFX®, Harrisburg, PA), and differences between sources were analyzed using the Mann-Whitney U test in IBM SPSS Statistics software, version 25 (IBM Corp., Armonk, NY) and R (v4.3.2, The R Core Team, R Foundation for Statistical Computing, Vienna, Austria). Results UpToDate had a higher word count and higher words per sentence than ChatGPT (both p < 0.05). ChatGPT had a lower absolute difficult-word count (p = 0.002) but a higher difficult-word percentage (p = 0.002). Differences in Flesch Reading Ease (FRE), Flesch-Kincaid Grade Level (FKGL), Simple Measure of Gobbledygook (SMOG), and sentence count were not statistically significant (all <i>p</i> > 0.05). Conclusions ChatGPT produced more concise content compared to UpToDate, but its higher proportion of difficult words may limit comprehension, highlighting the need to balance brevity with readability in AI-generated medical information.

Ähnliche Arbeiten