FIND: Fine-grained Information Density Guided Adaptive Retrieval-Augmented Generation for Disease Diagnosis

2025-02-20

AI Summary

Existing medical RAG systems lack dynamic assessment of diagnostic task difficulty, hindering the balance between accuracy and inference efficiency in clinical settings. This paper proposes an adaptive RAG framework tailored for clinical diagnosis: (1) a fine-grained information density prediction model enables interpretable, context-aware retrieval triggering; (2) a clinical-semantic-aware knowledge filtering module enhances retrieved evidence relevance; and (3) both components are integrated into a retrieval-augmented generation architecture. Experiments on three Chinese electronic health record datasets demonstrate that our method achieves up to 9.2% higher diagnostic accuracy and 37% lower response latency compared to state-of-the-art RAG baselines, significantly improving the accuracy–efficiency trade-off. The core contributions lie in a task-difficulty-driven adaptive retrieval mechanism and clinical-semantic-adapted knowledge filtering.

Technology Category

• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

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

Abstract

Retrieval-Augmented Large Language Models (LLMs), which integrate external knowledge into LLMs, have shown remarkable performance in various medical domains, including clinical diagnosis. However, existing RAG methods struggle to effectively assess task difficulty to make retrieval decisions, thereby failing to meet the clinical requirements for balancing efficiency and accuracy. So in this paper, we propose FIND ( extbf{F}ine-grained extbf{In}formation extbf{D}ensity Guided Adaptive RAG), a novel framework that improves the reliability of RAG in disease diagnosis scenarios. FIND incorporates a fine-grained adaptive control module to determine whether retrieval is necessary based on the information density of the input. By optimizing the retrieval process and implementing a knowledge filtering module, FIND ensures that the retrieval is better suited to clinical scenarios. Experiments on three Chinese electronic medical record datasets demonstrate that FIND significantly outperforms various baseline methods, highlighting its effectiveness in clinical diagnosis tasks.

Problem

• Improves retrieval decisions in disease diagnosis
• Balances efficiency and accuracy in clinical scenarios
• Enhances reliability of Retrieval-Augmented LLMs

Innovation

• Fine-grained adaptive control module
• Knowledge filtering module optimization
• Information density guided retrieval