TinyGate: A Confidence-Aware Lightweight Framework for Real-Time EEG Dementia Screening

View at publisher Summarise with VIRAT
Authors: Hyeonseo Jo, Injae Lee, Jin-Hwi Park, Young Chul Youn, Joonki Paik
Cited by: 0
References: 14

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Efficient and reliable EEG analysis is essential for real-time and home-based dementia screening. However, existing deep learning models remain too computationally heavy for low-power consumer devices and often perform redundant inference on easily classifiable signals. To address these limitations, we propose TinyGate, a lightweight and confidence-aware EEG framework designed for adaptive abnormality screening using only a few EEG channels. TinyGate dynamically determines whether an EEG segment can be confidently classified as normal or abnormal and provides direct feedback on whether further medical consultation is needed. For ambiguous cases, the system seamlessly transfers the data to a more advanced diagnostic backend for refined analysis. This selective inference approach significantly reduces redundant computation while maintaining diagnostic reliability. In general, TinyGate demonstrates a promising step toward real-time, scalable, and energy-efficient EEG screening for next-generation consumer healthcare devices.

Fields of Study & Subjects

EEG and Brain-Computer Interfaces ECG Monitoring and Analysis Emotion and Mood Recognition

Publication details

Journal / source
International Conference on Consumer Electronics
Pages
1–5
Publication date
3 February 2026
Language
EN