The assessment of the validity, safety, and utility of ChatGPT for patients with herniated lumbar disc: A preliminary study

Abstract

ChatGPT is perceived as a potential tool for patients diagnosed with herniated lumbar disc (HLD) to ask questions concerning desired information, with provision for necessary responses. In this preliminary study, we assessed the validity, safety, and utility of ChatGPT in patients with HLD. Two physicians specializing in the treatment of musculoskeletal disorders discussed and determined the 12 most frequently asked questions by patients with HLD in clinical practice. We used ChatGPT (version 4.0) to ask questions related to HLD. Each question was inputted into ChatGPT, and the responses were assessed by the 2 physicians. A Likert score was used to evaluate the validity, safety, and utility of the responses generated by ChatGPT. Each score for validity, safety, and utility was divided into 4 points, with a score of 4 indicating the most valid, safe, and useful answers and 1 point indicating the worst answers. Regarding validity, ChatGPT responses demonstrated 4 points for 9 questions (9/12, 75.0%) and 3 points for 3 questions (3/12, 25.0%). Regarding safety, ChatGPT scored 4 points for 11 questions (11/12, 91.7%) and 3 points for 1 question (1/12, 8.3%). Regarding utility, ChatGPT responses exhibited 4 points for 9 questions (9/12, 75.0%) and 3 points for 3 questions (3/12, 25.0%). ChatGPT demonstrates a tendency to offer relatively valid, safe, and useful information regarding HLD. However, users should exercise caution as ChatGPT may occasionally provide incomplete answers to some questions on HLD.