Quality of information provided by artificial intelligence for assigned female at birth patients undergoing gender affirming surgery

Abstract

Introduction. Artificial Intelligence has emerged as a transformative force across various industries, with a particularly profound impact on healthcare. It is well known that patients today increasingly turn to the internet searching for information about their medical conditions, utiliz-ing tools like AI-based chatbots. However, information from unverified sources can influence patients’ decisions regarding treatment options. This study aims to evaluate the quality of medical information provided by ChatGPT for Assigned Female At Birth (AFAB) patients considering gender-affirming surgery.Methods. Given the possibility that some patients might use ChatGPT as an information source for their medical conditions, specific ques-tions were posed to the chatbot in the same manner a patient inter-ested in gender-affirming surgery would. The quality of the information was assessed using the standardized EQIP scale. The survey involved 30 individuals: 15 plastic surgery residents and 15 non-healthcare pro-fessionals, with data collected in February 2023 and analyzed using SPSS Software version 28.0.Results. Separate surveys evaluated the quality of information provid-ed by ChatGPT regarding two primary procedures for AFAB patients undergoing gender-affirming surgery: phalloplasty and top surgery. The quality of the information was found to be adequate in both cases, with significant qualitative differences across the various survey sections. ChatGPT excelled in delivering information in a simple and accessible manner, earning high scores in the “Structured Data” area. However, the “Content Data” area, representing the completeness of information, was deemed sufficient. A significant deficiency was noted in the “Iden-tification Data” section, highlighting the absence of information about revisions, bibliographies, and the names of the entities or individuals providing content.Conclusions. ChatGPT demonstrated excellent capability in providing information in a straightforward and accessible manner, achieving high scores in the “Structured Data” area in both evaluations. The complete-ness of information, represented in the “Content Data” area, was con-sidered sufficient. However, a notable deficiency in the “Identification Data” section underscored the absence of details regarding revisions, bibliographies, and content authorship. Although the content score F.R. Grippaudo et al.2could be improved by adjusting the number and phrasing of questions, the lack of bibliography and source verification remains a significant limitation of this tool. ChatGPT offers advantages such as ease of communi-cation, privacy, anonymity, and overcoming language barriers; nonetheless, given its limitations, its role should always be seen as supplementary to that of the surgeon.