Authors: Xuanzhong Chen, Xiaohao Mao, Qihan Guo, Lun Wang, Shuyang Zhang, Ting Chen

Published on: February 09, 2024

Impact Score: 8.3

Arxiv code: Arxiv:2402.06341

Summary

• What is new: Introduction of RareBench, a benchmark for evaluating Large Language Models (LLMs) on rare diseases, and the largest open-source dataset on rare disease patients.
• Why this is important: Low clinical diagnosis rates for rare diseases due to the scarcity of experienced physicians and the complexity of differential diagnosis.
• What the research proposes: A dynamic few-shot prompt methodology that uses a comprehensive rare disease knowledge graph to enhance the diagnostic capabilities of LLMs.
• Results: Experimental findings highlight the superior diagnostic performance of GPT-4 over specialist physicians in the context of rare diseases.

Technical Details

Technological frameworks used: Dynamic few-shot prompting

Models used: GPT-4

Data used: The largest open-source dataset on rare disease patients

Potential Impact

Healthcare providers, telemedicine platforms, diagnostic software companies

Want to implement this idea in a business?

We have generated a startup concept here: DiagnoRare.