Pushing the P300-based brain-computer interface beyond 100 bpm: extending performance guided constraints into the...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• P300-based brain-computer interfaces (BCIs) traditionally use fixed-timing paradigms.
• Existing paradigms like the checkerboard paradigm (CBP) offer improvements over standard row/column methods.
• There is a need for more efficient and flexible BCI presentation strategies.

Purpose of the Study:

• Introduce and evaluate a novel asynchronous paradigm (ASP) for P300-based BCIs.
• Extend performance-guided constraints from spatial to temporal domains.
• Improve information transfer rates and overall BCI performance.

Main Methods:

• Developed temporal constraints to derive the ASP.
• Compared ASP performance against the checkerboard paradigm (CBP).
• Utilized stepwise linear discriminant analysis on 11 minutes of calibration data for online classification.

Main Results:

• The ASP demonstrated a 45% higher information transfer rate compared to the CBP.
• High accuracy was maintained in both paradigms (87% ASP, 93% CBP).
• A free spelling task achieved 100% accuracy with a practical throughput of 120 bits per minute.

Conclusions:

• The ASP represents a significant advancement in P300 speller technology, raising performance ceilings for BCIs.
• The study presents a novel and effective general strategy for organizing item presentation timing.
• This asynchronous approach has potential benefits for individuals with neuromuscular disabilities.