An Active RBSE Framework to Generate Optimal Stimulus Sequences in a BCI for Spelling

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Brain-computer interfaces (BCIs) using electroencephalography (EEG) aid individuals with severe motor and speech impairments.
• Current BCIs often use event-related potentials (ERPs) for typing, but performance can be limited by random character selection.
• Optimizing character presentation is key to enhancing typing speed and accuracy in BCIs.

Purpose of the Study:

• To introduce and evaluate the active recursive Bayesian state estimation (active-RBSE) framework for optimizing character selection in EEG-based BCIs.
• To enhance inference and sequence optimization for improved typing performance.
• To assess the impact of adaptive, user-specialized queries on BCI typing speed and accuracy.

Main Methods:

• Developed the active-RBSE framework for optimal subset selection based on a query function before character presentation.
• Compared active-RBSE against standard paradigms like matrix and random rapid serial visual presentation.
• Conducted simulation-based studies and real-time experiments with human participants to evaluate performance.

Main Results:

• The active-RBSE framework significantly enhanced online typing speed and accuracy in simulations.
• Real-time experiments with human participants confirmed improvements in both typing speed and accuracy.
• The framework demonstrated an adaptive, user-specialized approach to query selection.

Conclusions:

• The active-RBSE framework offers a significant advancement for EEG-based BCI typing systems.
• Optimized character selection through active-RBSE leads to better user interaction and communication.
• This approach holds promise for improving assistive technologies for individuals with severe communication impairments.