MedAide: Towards an Omni Medical Aide via Specialized LLM-based Multi-Agent Collaboration

2024-10-16

arXiv.org

AI Summary

Medical large language models (LLMs) face challenges in fusing heterogeneous, multi-intent information; suffer from severe hallucinations due to intent coupling; and lack personalized clinical decision-making capabilities. Method: This paper proposes a multi-agent collaborative framework for LLMs tailored to comprehensive healthcare scenarios. It introduces a novel fine-grained intent recognition mechanism based on intent prototype embeddings, integrated with query rewriting, semantic similarity matching, and dynamic agent scheduling to enable on-demand, specialized agent collaboration under composite intents. Reliability is enhanced via retrieval-augmented generation (RAG), context encoders, and medical knowledge-aligned fine-tuning. Results: Evaluated on four composite-intent medical benchmarks, the framework significantly outperforms state-of-the-art LLMs. Both automated evaluation and double-blind assessments by physicians from top-tier hospitals confirm substantial improvements in medical reasoning accuracy and diagnostic strategicity.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Multiagent Learning
• Natural Language Processing: (Large) Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Large Language Model (LLM)-driven interactive systems currently show potential promise in healthcare domains. Despite their remarkable capabilities, LLMs typically lack personalized recommendations and diagnosis analysis in sophisticated medical applications, causing hallucinations and performance bottlenecks. To address these challenges, this paper proposes MedAide, an LLM-based omni medical multi-agent collaboration framework for specialized healthcare services. Specifically, MedAide first performs query rewriting through retrieval-augmented generation to accomplish accurate medical intent understanding. Immediately, we devise a contextual encoder to obtain intent prototype embeddings, which are used to recognize fine-grained intents by similarity matching. According to the intent relevance, the activated agents collaborate effectively to provide integrated decision analysis. Extensive experiments are conducted on four medical benchmarks with composite intents. Experimental results from automated metrics and expert doctor evaluations show that MedAide outperforms current LLMs and improves their medical proficiency and strategic reasoning.

Problem

• Fusing heterogeneous medical information from diverse sources
• Reducing redundancy and coupling in complex medical intents
• Improving adaptive intent recognition in healthcare dialogues

Innovation

• LLM-based multi-agent collaboration for medical intent fusion
• Regularization-guided query decomposition with syntactic constraints
• Dynamic intent prototype matching for adaptive recognition