Vigilance Estimating in SSVEP-Based BCI Using Multimodal Signals.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) enable direct neural control of external devices.
• Monitoring user vigilance is crucial for effective and safe BCI operation.
• Existing vigilance estimation methods may lack accuracy or multimodal integration.

Purpose of the Study:

• To investigate vigilance changes in participants during a continuous BCI task.
• To develop and validate a multimodal method for estimating BCI user vigilance.
• To assess the efficacy of different physiological signals and machine learning models.

Main Methods:

• A high-speed 4-target BCI system using steady-state visual evoked potentials (SSVEP) was implemented.
• Simultaneous recording of electroencephalogram (EEG), electrooculogram (EOG), electrocardiography (ECG), and electrodermal activity (EDA) from 18 participants.
• Feature extraction, including differential entropy (DE), and application of support vector regression (SVR) for vigilance estimation.

Main Results:

• Differential entropy (DE) from EEG signals effectively reflected vigilance fluctuations.
• The multimodal approach integrating DE and EOG features into an SVR model outperformed other methods.
• The developed method demonstrated feasibility for real-time vigilance estimation in BCI users.

Conclusions:

• Multimodal physiological signal analysis, particularly integrating EEG and EOG, is a viable strategy for estimating BCI user vigilance.
• The proposed DE and EOG-based SVR model offers a promising solution for monitoring user attention.
• This research contributes to enhancing the safety and performance of BCI systems.