歯周病学における人工知能チャットボットが提供する情報の信頼性

Artificial intelligence chatbots, widely used in various fields including health care, are increasingly being adopted by patients seeking answers to their dental questions. This study compares the accuracy of chatbot responses to frequently asked questions in periodontology with those provided by expert periodontists. The aim of this study is to evaluate and compare the accuracy of answers given by ChatGPT-4, Google Gemini 1.5 Flash, ChatCPT-4.0o, Google Gemini 1.5 Pro to frequently asked questions in the field of periodontics. Frequently asked questions (n=10) were formulated by periodontologists and collected via ChatGPT-4 queries (n=10); each question was asked to the chatbot once. The responses (n=80) were independently assessed by 2 dentists and one experienced periodontologist using a modified Global Quality Score (GQS) on a 5-point Likert scale. Disagreements on scoring were resolved through evidence-based discussions. Kruskal-Wallis test was performed for the accuracy of responses between chatbots. The average GQS scores for Gemini 1.5 Flash, Gemini 1.5 Pro, ChatGPT-4, and ChatGPT-4.0o were 4.5, 4.8, 3.2, and 4.6, respectively. A significant difference was observed among the chatbots in terms of GQS scores (P<0.001). This significance was attributed to the differences between the following pairs: ChatGPT-4 and ChatGPT-4.0o (P<0.001), Gemini 1.5 Pro and ChatGPT-4 (P<0.001), and Gemini 1.5 and ChatGPT-4 (P<0.001). ChatGPT-4.0o, Google Gemini 1.5 Flash, and Google Gemini 1.5 Pro provided more accuracy information on topics related to periodontology compared with ChatGPT-4.