Fig. 2: Query pipeline of the LLMs.

The content of the question consists of two parts a prompt to ask LLMs to provide several possible diagnoses and one most likely diagnosis, and a detailed case report. Each case was individually queried four times.