Enhancing Medical Dialogue Generation through Knowledge Refinement and Dynamic Prompt Adjustment

2025-06-12

AI Summary

Existing medical dialogue systems suffer from inadequate precision in medical knowledge retrieval and suboptimal generation of personalized, clinically accurate responses. To address these limitations, we propose MedRef—a novel framework featuring a knowledge refinement mechanism and a dynamic prompt adaptation architecture. MedRef integrates a Triplet Filter for structured medical knowledge pruning and a Demo Selector for context-aware, patient-state-driven exemplar retrieval. Methodologically, it combines structured historical/evidence prompting, retrieval-augmented generation (RAG), and multi-stage knowledge filtering. Evaluated on the MedDG and KaMed benchmarks, MedRef achieves significant improvements in response quality—measured by BLEU (+4.2%) and ROUGE-L (+7.8%)—and medical entity accuracy, demonstrating enhanced clinical fidelity and reliability. This work advances medical dialogue systems by enabling adaptive, knowledge-grounded, and patient-centric response generation.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Natural Language Processing: Prompt Engineering / Prompting

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Medical dialogue systems (MDS) have emerged as crucial online platforms for enabling multi-turn, context-aware conversations with patients. However, existing MDS often struggle to (1) identify relevant medical knowledge and (2) generate personalized, medically accurate responses. To address these challenges, we propose MedRef, a novel MDS that incorporates knowledge refining and dynamic prompt adjustment. First, we employ a knowledge refining mechanism to filter out irrelevant medical data, improving predictions of critical medical entities in responses. Additionally, we design a comprehensive prompt structure that incorporates historical details and evident details. To enable real-time adaptability to diverse patient conditions, we implement two key modules, Triplet Filter and Demo Selector, providing appropriate knowledge and demonstrations equipped in the system prompt. Extensive experiments on MedDG and KaMed benchmarks show that MedRef outperforms state-of-the-art baselines in both generation quality and medical entity accuracy, underscoring its effectiveness and reliability for real-world healthcare applications.

Problem

• Improving accuracy of medical knowledge identification in dialogues
• Enhancing personalized and medically precise response generation
• Enabling real-time adaptation to diverse patient conditions

Innovation

• Knowledge refining mechanism filters irrelevant medical data
• Dynamic prompt adjustment with historical and evident details
• Triplet Filter and Demo Selector for real-time adaptability