Neurophysiological predictor of SMR-based BCI performance.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) enable computer control via brain activity, typically measured by electroencephalography (EEG).
• BCI success varies significantly among users, with approximately 20% experiencing 'BCI illiteracy'—insufficient control accuracy.
• Predicting BCI performance is crucial for understanding BCI illiteracy and optimizing user training.

Purpose of the Study:

• To identify a neurophysiological predictor of BCI performance.
• To develop a method for quantifying BCI predictor values from physiological data.
• To mitigate BCI illiteracy and improve user training efficiency.

Main Methods:

• A neurophysiological predictor was developed using a two-minute 'relax with eyes open' EEG recording.
• Laplacian EEG channels were utilized for data acquisition.
• The predictor was validated on a dataset of 80 BCI-naive participants during their initial session with the Berlin BCI (BBCI) system.

Main Results:

• A significant correlation (r=0.53) was found between the proposed neurophysiological predictor and BCI feedback performance.
• The predictor was determined from a brief EEG recording, suggesting practical applicability.
• The study utilized a large dataset (N=80) of participants new to BCI systems.

Conclusions:

• The proposed neurophysiological measure serves as a reliable predictor of BCI performance.
• This predictor can help identify individuals likely to experience BCI illiteracy.
• The findings pave the way for personalized BCI training and improved user outcomes.