RareAgents: Autonomous Multi-disciplinary Team for Rare Disease Diagnosis and Treatment

TL;DR

RareAgents introduces a patient-centered, autonomous MDT framework for rare disease diagnosis and treatment, integrating dynamic long-term memory and medical tools atop Llama-3.1 to coordinate a multidisciplinary team of 41 specialist pools. Across RareBench and MIMIC-IV-Ext-Rare, RareAgents with MDT achieves superior diagnostic accuracy and safer medication recommendations compared with domain-specific models and existing medical agents, with memory and tool modules driving key gains. The work also provides the MIMIC-IV-Ext-Rare dataset to facilitate future research and demonstrates the framework’s plug-and-play adaptability for real-world clinical workflows. These findings suggest MDT-guided, memory-augmented LLM systems can meaningfully augment complex precision medicine for rare diseases while highlighting the need for cautious deployment and multimodal data integration in clinical practice.

Abstract

Rare diseases, despite their low individual incidence, collectively impact around 300 million people worldwide due to the vast number of diseases. The involvement of multiple organs and systems, and the shortage of specialized doctors with relevant experience, make diagnosing and treating rare diseases more challenging than common diseases. Recently, agents powered by large language models (LLMs) have demonstrated notable applications across various domains. In the medical field, some agent methods have outperformed direct prompts in question-answering tasks from medical examinations. However, current agent frameworks are not well-adapted to real-world clinical scenarios, especially those involving the complex demands of rare diseases. To bridge this gap, we introduce RareAgents, the first LLM-driven multi-disciplinary team decision-support tool designed specifically for the complex clinical context of rare diseases. RareAgents integrates advanced Multidisciplinary Team (MDT) coordination, memory mechanisms, and medical tools utilization, leveraging Llama-3.1-8B/70B as the base model. Experimental results show that RareAgents outperforms state-of-the-art domain-specific models, GPT-4o, and current agent frameworks in diagnosis and treatment for rare diseases. Furthermore, we contribute a novel rare disease dataset, MIMIC-IV-Ext-Rare, to facilitate further research in this field.

Figures (24)

• Figure 1: A rare disease patient with multi-organ / multi-system abnormalities necessitates a multi-disciplinary team for comprehensive diagnosis and treatment.
• Figure 2: Overview of the RareAgents: For patients with multiple organ / multiple system abnormalities, the Attending Physician Agent selects specialists from a predefined pool to form an MDT, which reaches consensus through iterative discussions. Each physician agent is equipped with: a) Dynamic long-term memory to retrieve and update similar cases or prior records; and b) Tools to assist in diagnosis or treatment decisions.
• Figure 3: Agents Number in MDT.
• Figure 4: Diagnosis MDT settings.
• Figure 5: Treatment MDT settings.