DDO: Dual-Decision Optimization for LLM-Based Medical Consultation via Multi-Agent Collaboration

TL;DR

This paper introduces DDO, a dual-decision optimization framework that improves LLM-based medical consultation by separately optimizing symptom inquiry and disease diagnosis through multi-agent collaboration.

Contribution

It proposes a novel decoupled, multi-agent approach for LLM-based medical consultation, addressing the dual-task nature of symptom inquiry and diagnosis.

Findings

DDO outperforms existing LLM-based methods on real-world datasets.

Achieves competitive results with state-of-the-art generation models.

Demonstrates effectiveness in complex medical decision-making tasks.

Abstract

Large Language Models (LLMs) demonstrate strong generalization and reasoning abilities, making them well-suited for complex decision-making tasks such as medical consultation (MC). However, existing LLM-based methods often fail to capture the dual nature of MC, which entails two distinct sub-tasks: symptom inquiry, a sequential decision-making process, and disease diagnosis, a classification problem. This mismatch often results in ineffective symptom inquiry and unreliable disease diagnosis. To address this, we propose \textbf{DDO}, a novel LLM-based framework that performs \textbf{D}ual-\textbf{D}ecision \textbf{O}ptimization by decoupling the two sub-tasks and optimizing them with distinct objectives through a collaborative multi-agent workflow. Experiments on three real-world MC datasets show that DDO consistently outperforms existing LLM-based approaches and achieves competitive performance with state-of-the-art generation-based methods, demonstrating its effectiveness in the MC task. The code is available at https://github.com/zh-jia/DDO.