Evaluating the performance of large language models in health education for patients with ankylosing spondylitis/spondyloarthritis: a cross-sectional, single-blind study in China.

Abstract

To evaluate the potential of large language models (LLMs) in health education for patients with ankylosing spondylitis (AS)/spondyloarthritis (SpA), focusing on the accuracy of information transmission, patient acceptance and performance differences between different models. Cross-sectional, single-blind study. Multiple centres in China. 182 volunteers, including 4 rheumatologists and 178 patients with AS/SpA. Scientificity, precision and accessibility of the content of the answers provided by LLMs; patient acceptance of the answers. LLMs performed well in terms of scientificity, precision and accessibility, with ChatGPT-4o and Kimi models outperforming traditional guidelines. Most patients with AS/SpA showed a higher level of understanding and acceptance of the responses from LLMs. LLMs have significant potential in medical knowledge transmission and patient education, making them promising tools for future medical practice.