Related Experiment Video
Updated: Apr 21, 2026
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
5.8K
Attention-based multi-feature fusion neuromarker for EEG-driven stress classification in learners
Saliha Ejaz1, Soyiba Javed1, Imran Shafi1,2
1Department of Computer and Software Engineering, College of Electrical and Mechanical Engineering, National University of Sciences and Technology (NUST), Islamabad 44000, Pakistan.
Summary
This study introduces a novel method for detecting mental stress using electroencephalogram (EEG) signals. By combining multiple brain signal features, the approach achieves high accuracy in classifying stress, offering a more objective measure than traditional self-assessments.
Area of Science:
- Neuroscience
- Computational Psychiatry
- Biomedical Engineering
Background:
- Student mental stress is a growing concern, impacting academic performance and well-being.
- Traditional self-assessment for stress is subjective and prone to errors.
- Electroencephalogram (EEG) offers objective insights into brain states.
Purpose of the Study:
- To develop an objective, multi-feature fusion model for accurate stress classification using EEG signals.
- To explore novel neuromarkers for stress by integrating spatial and connectivity features.
- To investigate the utility of microstate analysis for stress-related brain network changes.
Main Methods:
- A feature-fusion model combining spatial (Microstates - MS) and connectivity (Transfer Entropy - TE, Granger Causality - GC) EEG features.
- Incorporation of attention fusion to enhance discriminative features and mitigate individual modality limitations.
- Classification using Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), and Multilayer Perceptron (MLP).
Main Results:
- The proposed attention-fusion multi-feature model achieved high classification accuracies: 95.47% (SVM), 98.91% (LDA), and 83.49% (MLP).
- Novel microstate topomaps for stress-induced brain networks were identified.
- Validation confirmed the robustness of the multi-feature fusion approach over individual or binary feature combinations.
Conclusions:
- The developed feature-fusion model provides a robust neuromarker for stress classification, capturing complex brain dynamics.
- This objective approach offers a significant advancement over subjective stress assessment methods.
- The findings open new avenues for understanding and managing student mental health through neurophysiological markers.