Kshitij Sharad Jadhav

Few shot chain-of-thought driven reasoning to prompt LLMs for open-ended medical question answering

In this paper, we propose a modified version of the MedQA-USMLE dataset, named MEDQA-OPEN, which contains open-ended medical questions without options to mimic clinical scenarios, along with clinician-approved reasoned answers. Additionally, we implement a prompt driven by Chain of Thought (CoT) rea…