Answers are grounded in this paper and cite their sources.
Novel framework improves classification accuracy for Alzheimer’s disease and mild cognitive impairment using EEG data.
The electroencephalography (EEG)-based approach provides a promising low-cost and non-invasive approach to the early detection of pathological cognitive decline. However, current studies predominantly utilize EEGs from resting state (rsEEG) or task-state (task EEG), posing challenges to classification performances due to the unconstrainted nature of mind wandering during resting state or the inherent inter-participant variability from task execution. To address these limitations, this study proposes a novel feature extraction framework, working memory task-induced EEG response (WM-TIER), which adjusts task EEG features by rsEEG features and leverages the often-overlooked inter-state changes of EEGs. We recorded EEGs from 21 AD individuals, 24 MCI individuals, and 27 healthy controls (HC) during both resting and working memory task conditions. We then compared the classification performance of WM-TIER to the conventional rsEEG or task EEG framework. For each framework, three feature types were examined: relative power, spectral coherence, and filter-bank phase lag index (FB-PLI). Our results indicated that FB-PLI-based WM-TIER features provide (1) better AD/MCI versus HC classification accuracy than rsEEG and task EEG frameworks and (2) high accuracy for three-class classification of AD vs. MCI vs. HC. These findings suggest that the EEG-based rest-to-task state transition can be an effective neural marker for the early detection of pathological cognitive decline.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: