Background/aims This was a cross-sectional comparative study. We evaluated the ability of three large language models (LLMs) (ChatGPT-3.5, ChatGPT-4, and Google Bard) to generate novel patient ...