Online BCI Typing using Language Model Classifiers by ALS Patients in their Homes

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Brain-computer interfaces (BCIs) offer communication pathways for individuals with severe motor impairments, such as amyotrophic lateral sclerosis (ALS).
• Previous research indicates that integrating linguistic information into P300 speller signal classification can boost performance.
• However, these advancements have primarily been validated in laboratory settings with healthy participants.

Purpose of the Study:

• To evaluate the real-world functionality of a P300 speller system augmented with language models for ALS patients.
• To assess the system's performance in a home environment, reflecting daily living conditions.

Main Methods:

• Six individuals diagnosed with ALS, exhibiting varying functional levels, participated in the study.
• Offline performance metrics were analyzed with and without the language model.
• Online communication sessions were conducted to measure typing speed and accuracy.

Main Results:

• All ALS participants demonstrated improved offline performance when utilizing the language model.
• Five participants achieved online typing speeds of at least six characters per minute with accuracy exceeding 84%.

Conclusions:

• The performance enhancements observed with language models in P300 spellers are transferable to the ALS patient population.
• This study supports the potential of language model-integrated P300 spellers as effective assistive communication devices for individuals with ALS in their homes.