KERAP: A Knowledge-Enhanced Reasoning Approach for Accurate Zero-shot Diagnosis Prediction Using Multi-agent LLMs

2025-07-03

AI Summary

To address critical challenges in zero-shot medical diagnosis—including poor model generalization, frequent hallucinations in large language models (LLMs), and insufficient structured reasoning and interpretability—this paper proposes a knowledge graph (KG)-enhanced multi-agent collaborative reasoning framework. The method aligns clinical phenotypes with KG ontologies via attribute mapping, integrates dynamic KG-based knowledge retrieval, and employs iterative multi-turn optimization to enable structured, verifiable diagnostic reasoning. It synergistically combines LLMs’ semantic comprehension, KGs’ logical constraints, and multi-agent specialization. Evaluated on multiple zero-shot diagnostic benchmarks, our approach achieves significant improvements in accuracy (+12.7%) and result reliability, effectively mitigates hallucinations, and generates clinically compliant, structured diagnostic reports. The framework ensures strong interpretability and demonstrates practical deployability.

Technology Category

• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Machine Learning: Large Multimodal Models (LMMs)
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Medical diagnosis prediction plays a critical role in disease detection and personalized healthcare. While machine learning (ML) models have been widely adopted for this task, their reliance on supervised training limits their ability to generalize to unseen cases, particularly given the high cost of acquiring large, labeled datasets. Large language models (LLMs) have shown promise in leveraging language abilities and biomedical knowledge for diagnosis prediction. However, they often suffer from hallucinations, lack structured medical reasoning, and produce useless outputs. To address these challenges, we propose KERAP, a knowledge graph (KG)-enhanced reasoning approach that improves LLM-based diagnosis prediction through a multi-agent architecture. Our framework consists of a linkage agent for attribute mapping, a retrieval agent for structured knowledge extraction, and a prediction agent that iteratively refines diagnosis predictions. Experimental results demonstrate that KERAP enhances diagnostic reliability efficiently, offering a scalable and interpretable solution for zero-shot medical diagnosis prediction.

Problem

• Improves zero-shot diagnosis prediction using multi-agent LLMs
• Addresses hallucinations and lack of structured medical reasoning in LLMs
• Enhances diagnostic reliability with knowledge graph-enhanced reasoning

Innovation

• Knowledge graph-enhanced reasoning for diagnosis
• Multi-agent architecture improves LLM predictions
• Zero-shot medical diagnosis with structured knowledge