Interpretable feature-transformer framework for cross-subject MCI detection using nonlinear dynamical and graph-theoretic EEG features.
The feature-based Transformer achieved a high performance of 97.04% in distinguishing MCI from healthy controls, indicating potential for advanced EEG analysis in early detection.
Where it sits
this study against the rest of the lixisenatide corpusSummary and findings
This study investigated the effectiveness of entropy- and graph-based EEG features for distinguishing Mild Cognitive Impairment (MCI) from healthy controls (HC) using two modeling approaches. The dataset consisted of resting-state EEG recordings from 183 participants, with 127 classified as HC and 56 as MCI. The feature-based Transformer achieved a performance of 97.04% ± 0.72.
Abstract
Early and accurate detection of Mild Cognitive Impairment (MCI) is essential for preventing progression toward Alzheimer's disease (AD). In this cross-subject study, we investigate the effectiveness of entropy- and graph-based EEG features for distinguishing MCI from healthy controls (HC), using two modeling approaches: (1) a Transformer network applied to the engineered feature set, and (2) an EEGNet model trained on the same feature representation for comparison. The dataset consists of resting-state, eyes-closed EEG recordings from 183 participants (127 HC, 56 MCI), collected using a 20-channel STAT<sup>™</sup> X24 wireless system and segmented into 3-second epochs. EEG data underwent standard preprocessing, including band-pass filtering, downsampling, normalization, and class-balancing augmentation applied to the minority class. From each channel, nonlinear dynamical measures (e.g., sample and fuzzy entropy, Higuchi fractal dimension, Lyapunov exponent) and graph-theoretic connectivity descriptors derived from coherence matrices across five frequency bands were extracted, yielding a structured 19[Formula: see text]77 feature representation. The feature-based Transformer achieved the best performance (97.04% ± 0.72), outperforming the feature-based EEGNet baseline and highlighting the benefits of combining rich handcrafted features with attention-based modeling. SHAP (SHapley Additive exPlanations) analysis provided global and local interpretability, revealing the most influential nonlinear and connectivity features as well as the EEG channels contributing most to classification. Overall, these results demonstrate the effectiveness of feature-Transformer integration and support the potential of interpretable feature-driven deep learning models for early MCI detection.