An Agentic System for Rare Disease Diagnosis with Traceable Reasoning

Summary

This paper talks about DeepRare, an AI system that helps doctors diagnose rare diseases more accurately by using a large language model to analyze different types of medical information.

What's the problem?

The problem is that rare diseases are hard to diagnose because they are very different from each other, occur rarely, and most doctors aren’t familiar with them, which causes delays in getting the right diagnosis.

What's the solution?

The researchers created DeepRare, which uses an intelligent system combining many specialized tools and up-to-date medical knowledge sources. It processes a wide range of patient information like written descriptions, genetic data, and clinical features, and it generates possible diagnoses along with clear explanations that link each conclusion to trusted medical evidence. The system also checks and revises its guesses to avoid mistakes.

Why it matters?

This matters because quicker and more accurate diagnosis of rare diseases can help patients get the right treatments sooner, reduce misdiagnoses, and support doctors by providing reliable and explainable AI assistance.