Enhancing the Performance of Event-Related Potential-Based Brain-Computer Interfaces Under Cognitive Distraction: A...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) utilizing event-related potentials (ERPs) typically demand high user attention.
• Real-world BCI applications often involve multitasking and cognitive distractions, which significantly impair ERP amplitudes and overall BCI performance.
• Existing BCI models struggle to maintain accuracy in distracting environments.

Purpose of the Study:

• To develop and validate a novel multiwindow adaptive model for visual ERP-based BCIs.
• To mitigate the performance degradation caused by cognitive distraction in real-world BCI scenarios.
• To enhance the robustness and practical applicability of ERP-based BCIs.

Main Methods:

• The proposed approach segments poststimulus intervals into multiple overlapping windows.
• Each window employs dedicated spatial filters and classifiers that are continuously updated via adaptive semi-supervised learning.
• The model was tested using offline experiments on a dataset collected during concurrent speaking and validated through online experiments.

Main Results:

• Offline experiments demonstrated that the multiwindow adaptive model significantly outperformed single-window and fixed models, achieving 91.08% accuracy.
• Online experiments confirmed the model's effectiveness in restoring BCI performance under cognitive distraction, reaching 93.20% accuracy.
• The results indicate substantial improvements in BCI performance despite challenging, distracting conditions.

Conclusions:

• The multiwindow adaptive model offers a practical solution for enhancing ERP-based BCIs in real-world, distracting environments.
• Temporally tailored feature extraction and continuous adaptation are crucial for robust BCI performance.
• This approach enables reliable BCI operation even when users are multitasking or experiencing cognitive load.