EEG-MedRAG: Enhancing EEG-based Clinical Decision-Making via Hierarchical Hypergraph Retrieval-Augmented Generation

TL;DR

This work tackles the challenge of making EEG-based clinical interpretation reliable amid large, heterogeneous data by introducing EEG-MedRAG, a three-layer hypergraph retrieval-augmented generation framework. It unifies domain knowledge, patient cases, and EEG waveforms into a traversable $n$-ary structure and pairs it with a semantic-temporal retrieval pipeline to ground LLM reasoning in concrete evidence. A cross-disease EEG QA benchmark spanning seven disorders and five clinical roles enables systematic evaluation of generalization and role-aware reasoning. Experimental results show consistent gains over competitive baselines in accuracy and retrieval quality, with ablations confirming the value of $n$-ary modeling and EEG-grounded fusion for clinically meaningful, interpretable generation.

Abstract

With the widespread application of electroencephalography (EEG) in neuroscience and clinical practice, efficiently retrieving and semantically interpreting large-scale, multi-source, heterogeneous EEG data has become a pressing challenge. We propose EEG-MedRAG, a three-layer hypergraph-based retrieval-augmented generation framework that unifies EEG domain knowledge, individual patient cases, and a large-scale repository into a traversable n-ary relational hypergraph, enabling joint semantic-temporal retrieval and causal-chain diagnostic generation. Concurrently, we introduce the first cross-disease, cross-role EEG clinical QA benchmark, spanning seven disorders and five authentic clinical perspectives. This benchmark allows systematic evaluation of disease-agnostic generalization and role-aware contextual understanding. Experiments show that EEG-MedRAG significantly outperforms TimeRAG and HyperGraphRAG in answer accuracy and retrieval, highlighting its strong potential for real-world clinical decision support. Our data and code are publicly available at https://github.com/yi9206413-boop/EEG-MedRAG.

Figures (8)

• Figure 1: EEG-MedRAG uses a hierarchical hypergraph to integrate EEG signals, patient records, and domain knowledge for improved clinical reasoning.
• Figure 2: The EEG-MedRAG benchmark covers seven EEG-related neurological disorders and supports five clinical roles: doctors, patients, researchers, hospital interns, and nurses, enabling tailored clinical reasoning and diagnostic support.
• Figure 3: An overview of EEG-MedRAG, which constructs a three-layer hypergraph from EEG domain knowledge, patient-specific data, and EEG waveforms, retrieves semantic-temporal information, and generates precise, clinical responses.
• Figure 4: An overview of EEG-MedRAG’s semantic-temporal EEG retrieval process, which calculates DTW distances between patient EEG signals and database waveforms to retrieve top-K similar EEG segments for clinical decision support.
• Figure 5: Bar charts comparing the Exact Match (EM) scores of EEG-MedRAG and baseline methods across five clinical roles under different LLM configurations: (a) GPT-4o-mini, (b) Deepseek-r1, and (c) Gemini-2.5-flash.