To assess the reliability and readability of information regarding acute cholecystitis produced by AI models ChatGPT®, Gemini®, and Perplexity® in English and Spanish.
Approach:
Study Design: A comparative cross-sectional study was conducted from April 1 to April 30, 2025, adhering to STROBE guidelines.
Sample and Questions: Seven standardized clinical questions were formulated, covering critical elements of acute cholecystitis, resulting in 42 responses from three AI models in two languages.
Response Generation: Responses were generated using ChatGPT-3.5®, Gemini®, and Perplexity AI® without premium features, focusing on initial outputs.
Response Evaluation: Two physicians evaluated the responses for reliability using a validated scale and assessed readability with specific formulas for English and Spanish.
Statistical Analysis: The Shapiro–Wilk test was used to assess data normality, with quantitative variables presented as medians and categorical variables as frequencies.
Key Findings:
AI-generated content on acute cholecystitis varies in reliability and readability.
Responses were evaluated based on completeness, correctness, and clarity.
Readability scores indicated varying levels of comprehensibility for patients.
Limitations:
The study did not conduct a formal sample size calculation.
Responses were evaluated by two physicians, which may introduce subjective bias.
The study focused only on initial responses from AI models, potentially overlooking follow-up clarifications.
by Mabel Lucero Olarte Jurado, Natalia Pimiento Blanco, María José Prieto Otero, Laura Sofía Rivera González, Miguel Enrique Ochoa Vera, Jaime Fernando Barajas, Cristian Orlando Porras Bueno
Patients with lower nonischemic resting full-cycle ratio values had more repeat revascularizations despite similar 2-year major cardiovascular event rates.
