SEMA-RAG: A Self-Evolving Multi-Agent Retrieval-Augmented Generation Framework for Medical Reasoning

2026-05-16

AI Summary

Traditional single-turn static retrieval-augmented generation (RAG) struggles to align with the multi-stage nature of clinical reasoning, often resulting in insufficient semantic understanding and weak evidence chain construction. This work proposes a self-evolving multi-agent RAG framework that, for the first time, decomposes clinical reasoning into three specialized agent roles: explanation, exploration, and arbitration. By integrating semantic-driven question parsing, dynamic multi-turn retrieval, and an evidence adjudication mechanism, the framework enables synergistic iteration between retrieval and reasoning. Extensive experiments across five medical question-answering benchmarks and five large language models demonstrate that the proposed method improves average accuracy by 6.46 percentage points over the strongest baseline.

Abstract

Retrieval-Augmented Generation (RAG) is widely employed to mitigate risks such as hallucinations and knowledge obsolescence in medical question answering, yet its predominantly single-round, static retrieval paradigm misaligns with the multi-stage process of clinical reasoning. This compressed workflow induces two structural deficiencies: question-to-query translation often lacks clinically grounded semantic interpretation, and retrieval lacks iterative sufficiency feedback, making it difficult to form reliable evidence chains. We argue that both issues stem from a deeper cause: overloading a single reasoning chain with heterogeneous tasks of interpretation, exploration, and adjudication. The remedy is to reconstruct the workflow via task decoupling and dynamic multi-round exploration. To this end, we propose SEMA-RAG, a Self-Evolving Multi-Agent RAG framework for medical question answering, which assigns these roles to three specialist agents: the Interpreter Agent for clinical schema interpretation, the Explorer Agent for sufficiency-driven self-evolving retrieval, and the Arbiter Agent for evidence adjudication and answer selection. Across five benchmarks and five LLM backbones, SEMA-RAG improves the strongest baseline by +6.46 accuracy points on average, measured per backbone.

Problem

• Retrieval-Augmented Generation
• medical reasoning
• clinical question answering
• evidence chain
• multi-round retrieval

Innovation

• Multi-Agent RAG
• Self-Evolving Retrieval
• Medical Reasoning
• Task Decoupling
• Iterative Evidence Chain