Transforming Brainwaves into Language: EEG Microstates Meets Text Embedding Models for Dementia Detection

Quoc-Toan Nguyen, Linh Le, Xuan-The Tran, Dorothy Bai, Nghia Duong-Trung, Thomas Do, Chin-teng Lin

Keywords: Text Embedding Model, EEG, Brain-Computer Interface, Dementia, Alzheimer's, AI in Healthcare

TL;DR: We propose a scalable, channel-independent framework that models EEG microstates as text-like sequences for deep learning-based Alzheimer’s detection (accuracy: 0.9431), enabling low-cost, accessible screening in resource-limited settings.

Abstract: Early detection of dementia, particularly Alzheimer's Disease (AD), its most prevalent form, is critical for slowing disease progression and improving quality of life through timely intervention. This study proposes a novel, scalable, and channel-independent approach that leverages EEG microstates, which are symbolic, linguistics-like representations of brain activity. It is processed with advanced text embedding and time-series deep learning techniques. Developed on EEG data from 1001 participants across multiple countries, the proposed method achieves a high accuracy of 0.9431 in AD detection. The proposed approach enhances generalisability and facilitates deployment in diverse, resource-limited settings by removing the need for fixed EEG channel configurations and expensive modalities. Its compatibility with low-cost EEG devices eliminates the requirement for distinct models, thereby reducing implementation costs and enabling scalable, accessible AD detection in underserved communities.

Archival Status: Archival

Paper Length: Long Paper (up to 8 pages of content)

Submission Number: 41