Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

EEGAgent: A Unified Framework for Automated EEG Analysis Using Large Language Models

Sha Zhao, Mingyi Peng, Haiteng Jiang, Tao Li, Shijian Li, Gang Pan

TL;DR

EEG analysis currently suffers from task-specific models that struggle with multi-task, continuous reasoning in real-world clinical settings. The authors introduce EEGAgent, a unified, LLM-driven framework that orchestrates a toolbox of parametric and non-parametric tools across perception, exploration, detection, interaction, and reporting. By leveraging an LLM planner, domain knowledge, and modular tools, EEGAgent enables context-aware, multi-task EEG analysis with interpretable outputs and automated report generation. Evaluations on TUAB, TUEV, and TUSL datasets demonstrate capabilities in perceptual understanding, coarse-to-fine event localization, and clinically structured reporting, highlighting potential for automated EEG interpretation and decision support in clinical workflows.

Abstract

Scalable and generalizable analysis of brain activity is essential for advancing both clinical diagnostics and cognitive research. Electroencephalography (EEG), a non-invasive modality with high temporal resolution, has been widely used for brain states analysis. However, most existing EEG models are usually tailored for individual specific tasks, limiting their utility in realistic scenarios where EEG analysis often involves multi-task and continuous reasoning. In this work, we introduce EEGAgent, a general-purpose framework that leverages large language models (LLMs) to schedule and plan multiple tools to automatically complete EEG-related tasks. EEGAgent is capable of performing the key functions: EEG basic information perception, spatiotemporal EEG exploration, EEG event detection, interaction with users, and EEG report generation. To realize these capabilities, we design a toolbox composed of different tools for EEG preprocessing, feature extraction, event detection, etc. These capabilities were evaluated on public datasets, and our EEGAgent can support flexible and interpretable EEG analysis, highlighting its potential for real-world clinical applications.

EEGAgent: A Unified Framework for Automated EEG Analysis Using Large Language Models

TL;DR

EEG analysis currently suffers from task-specific models that struggle with multi-task, continuous reasoning in real-world clinical settings. The authors introduce EEGAgent, a unified, LLM-driven framework that orchestrates a toolbox of parametric and non-parametric tools across perception, exploration, detection, interaction, and reporting. By leveraging an LLM planner, domain knowledge, and modular tools, EEGAgent enables context-aware, multi-task EEG analysis with interpretable outputs and automated report generation. Evaluations on TUAB, TUEV, and TUSL datasets demonstrate capabilities in perceptual understanding, coarse-to-fine event localization, and clinically structured reporting, highlighting potential for automated EEG interpretation and decision support in clinical workflows.

Abstract

Scalable and generalizable analysis of brain activity is essential for advancing both clinical diagnostics and cognitive research. Electroencephalography (EEG), a non-invasive modality with high temporal resolution, has been widely used for brain states analysis. However, most existing EEG models are usually tailored for individual specific tasks, limiting their utility in realistic scenarios where EEG analysis often involves multi-task and continuous reasoning. In this work, we introduce EEGAgent, a general-purpose framework that leverages large language models (LLMs) to schedule and plan multiple tools to automatically complete EEG-related tasks. EEGAgent is capable of performing the key functions: EEG basic information perception, spatiotemporal EEG exploration, EEG event detection, interaction with users, and EEG report generation. To realize these capabilities, we design a toolbox composed of different tools for EEG preprocessing, feature extraction, event detection, etc. These capabilities were evaluated on public datasets, and our EEGAgent can support flexible and interpretable EEG analysis, highlighting its potential for real-world clinical applications.

Paper Structure

This paper contains 18 sections, 5 equations, 5 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related works
  3. EEG Analysis
  4. LLM-Enhanced Agent Framework
  5. Method
  6. EEGAgent Architecture
  7. Toolbox Design for Capability Support
  8. EEG Perception Ability: Comprehending the fundamental information of EEG signals
  9. EEG Exploration Ability: Analyzing EEG segments from multiple perspectives
  10. EEG Detection Ability: Recognizing Targeted Spatiotemporal EEG Event
  11. Interaction and Reporting Ability: Interacting with Users and Generating Clinical EEG Reports
  12. Experiments
  13. Dataset
  14. EEG Perception Experiment
  15. EEG Segment Exploration Experiment
  16. ...and 3 more sections

Figures (5)

  • Figure 1: EEGAgent framework
  • Figure 2: The EEGAgent perceives the environment constructed from EEG data. Solid lines represent processes visible to the user.
  • Figure 3: Exploration of EEG segments by the EEGAgent. Solid lines represent processes visible to the user.
  • Figure 4: Detection of Epileptic Discharges by the EEGAgent. Solid lines represent processes visible to the user.
  • Figure 5: EEG report generated by the EEGAgent