Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Medical Graph RAG: Towards Safe Medical Large Language Model via Graph Retrieval-Augmented Generation

Junde Wu, Jiayuan Zhu, Yunli Qi, Jingkun Chen, Min Xu, Filippo Menolascina, Vicente Grau

TL;DR

Medical Graph RAG (MedGraphRAG) tackles safety and credibility challenges in medical LLMs by integrating a Triple Graph Construction with a hierarchical U-Retrieval pipeline to produce evidence-backed, source-grounded responses. The method constructs a multilayer knowledge graph anchored in RAG data, medical literature, and controlled vocabularies, and retrieves via top-down tag summaries with bottom-up refinement to balance global context and retrieval efficiency. Empirical results across nine medical QA benchmarks and two health fact-check datasets show MedGraphRAG achieving state-of-the-art performance on large models and notable gains on smaller models, with human clinicians rating its outputs as more credible and easier to understand. The work demonstrates practical significance for clinical decision support and health information dissemination, providing open-source code and releasing a dataset for further validation.

Abstract

We introduce a novel graph-based Retrieval-Augmented Generation (RAG) framework specifically designed for the medical domain, called \textbf{MedGraphRAG}, aimed at enhancing Large Language Model (LLM) capabilities for generating evidence-based medical responses, thereby improving safety and reliability when handling private medical data. Graph-based RAG (GraphRAG) leverages LLMs to organize RAG data into graphs, showing strong potential for gaining holistic insights from long-form documents. However, its standard implementation is overly complex for general use and lacks the ability to generate evidence-based responses, limiting its effectiveness in the medical field. To extend the capabilities of GraphRAG to the medical domain, we propose unique Triple Graph Construction and U-Retrieval techniques over it. In our graph construction, we create a triple-linked structure that connects user documents to credible medical sources and controlled vocabularies. In the retrieval process, we propose U-Retrieval which combines Top-down Precise Retrieval with Bottom-up Response Refinement to balance global context awareness with precise indexing. These effort enable both source information retrieval and comprehensive response generation. Our approach is validated on 9 medical Q\&A benchmarks, 2 health fact-checking benchmarks, and one collected dataset testing long-form generation. The results show that MedGraphRAG consistently outperforms state-of-the-art models across all benchmarks, while also ensuring that responses include credible source documentation and definitions. Our code is released at: https://github.com/MedicineToken/Medical-Graph-RAG.

Medical Graph RAG: Towards Safe Medical Large Language Model via Graph Retrieval-Augmented Generation

TL;DR

Medical Graph RAG (MedGraphRAG) tackles safety and credibility challenges in medical LLMs by integrating a Triple Graph Construction with a hierarchical U-Retrieval pipeline to produce evidence-backed, source-grounded responses. The method constructs a multilayer knowledge graph anchored in RAG data, medical literature, and controlled vocabularies, and retrieves via top-down tag summaries with bottom-up refinement to balance global context and retrieval efficiency. Empirical results across nine medical QA benchmarks and two health fact-check datasets show MedGraphRAG achieving state-of-the-art performance on large models and notable gains on smaller models, with human clinicians rating its outputs as more credible and easier to understand. The work demonstrates practical significance for clinical decision support and health information dissemination, providing open-source code and releasing a dataset for further validation.

Abstract

We introduce a novel graph-based Retrieval-Augmented Generation (RAG) framework specifically designed for the medical domain, called \textbf{MedGraphRAG}, aimed at enhancing Large Language Model (LLM) capabilities for generating evidence-based medical responses, thereby improving safety and reliability when handling private medical data. Graph-based RAG (GraphRAG) leverages LLMs to organize RAG data into graphs, showing strong potential for gaining holistic insights from long-form documents. However, its standard implementation is overly complex for general use and lacks the ability to generate evidence-based responses, limiting its effectiveness in the medical field. To extend the capabilities of GraphRAG to the medical domain, we propose unique Triple Graph Construction and U-Retrieval techniques over it. In our graph construction, we create a triple-linked structure that connects user documents to credible medical sources and controlled vocabularies. In the retrieval process, we propose U-Retrieval which combines Top-down Precise Retrieval with Bottom-up Response Refinement to balance global context awareness with precise indexing. These effort enable both source information retrieval and comprehensive response generation. Our approach is validated on 9 medical Q\&A benchmarks, 2 health fact-checking benchmarks, and one collected dataset testing long-form generation. The results show that MedGraphRAG consistently outperforms state-of-the-art models across all benchmarks, while also ensuring that responses include credible source documentation and definitions. Our code is released at: https://github.com/MedicineToken/Medical-Graph-RAG.
Paper Structure (29 sections, 4 equations, 3 figures, 3 tables)

This paper contains 29 sections, 4 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Medical Graph Construction
  4. Step1: Semantic Document Chunking
  5. Step2: Entities Extraction
  6. Step3: Triple Linking
  7. Step4: Relationship Linking
  8. Step5: Tag the Graphs
  9. Step6: U-Retrieval
  10. Experiment
  11. Dataset
  12. RAG data
  13. Repository data
  14. Test Data
  15. Experiment Setting
  16. ...and 14 more sections

Figures (3)

  • Figure 1: The overall workflow of MedGraphRAG.
  • Figure 2: Example responses from GraphRAG and MedGraphRAG, with abstracted graphs. MedGraphRAG provides more detailed explanations and more complex reasoning with evidences.
  • Figure 3: Impact of Repository Data on RAG, GraphRAG, and MedGraphRAG with GPT-4. Line chart: performance with incremental data inclusion; bar chart: performance with individual data inclusion.