GPT-doctor: Customizing Large Language Models for Medical Consultation

This paper proposes a novel framework to customize LLMs for general business contexts that aims to achieve three fundamental objectives simultaneously: (1) aligning conversational patterns, (2) integrating in-depth domain knowledge, and (3) embodying the soft skills and core principles. We design methodologies to combine domain-specific theory with Supervised Fine Tuning (SFT) in LLMs. We instantiate our proposed framework in the context of medical consultation and demonstrate the feasibility of our proposed framework using online experiments as well as evaluation by domain experts. Results show that our framework significantly narrows the gap between LLMs and human doctors, significantly moving closer to human-level performance. We demonstrate our model's practical use in a lab experiment by deploying it on an online consultation platform to support doctors. It helps doctors by providing initial responses in co-piloting fashion. Result shows that it boosts productivity and decreases cognitive load, proving its effectiveness for widespread use.