Artificial intelligence for knee osteoarthritis education: a comparative evaluation of ChatGPT and Gemini across Unani-inclusive patient domains

This study aims to compare the performance of two large language models (ChatGPT and Google Gemini) in generating patient education content for knee osteoarthritis, with specific emphasis on Unani medicine perspectives. The study evaluates AI-generated responses across 36 questions organized into 6 patient-relevant domains: (1) Understanding the condition, (2) Why I got this disease, (3) Treatment options, (4) Pain and functional management, (5) Diet, weight, and lifestyle, and (6) Safety, prognosis, and information needs. Three blinded expert reviewers will rate each AI-generated response on five dimensions: accuracy, completeness, clarity, safety, and trustworthiness using a standardized 5-point Likert scale. This comparative evaluation will assess whether AI chatbots can provide reliable, culturally-sensitive health information that incorporates traditional Unani therapeutic approaches alongside conventional medical knowledge for knee osteoarthritis patient education.