Overview of ChatBCI-Assist, integrating a P300-based BCI, a GUI, and a locally fine-tuned LLM to enable efficient communication. Keys on the GUI are selected by BCI by detecting P300 responses from the EEG signals using a subject-specific calibrated SWLDA classifier.
Invention Summary:
Neurodegenerative diseases such as Amyotrophic Lateral Sclerosis (ALS) can progressively impair speech and motor function, limiting an individual's ability to communicate through conventional methods such as speaking, writing, or typing. Brain-computer interface (BCI) technologies, including P300 speller systems, have emerged as promising augmentative and alternative communication (AAC) tools by enabling users to select characters using brain signals. However, traditional P300 spellers often require users to select characters one at a time, resulting in slow communication rates and substantial cognitive effort. There remains a need for more efficient and intelligent BCI communication systems that can improve typing speed, reduce user workload, and better capture intended messages.
Rutgers researchers have developed ChatBCI, an intelligent P300 speller brain-computer interface that integrates large language models with EEG-based intent detection to enhance communication for individuals with severe motor impairments. The technology combines brain-signal-driven selection with advanced language prediction capabilities to generate contextually relevant text and streamline message composition. The system incorporates adaptive interface features, including predictive text assistance and confidence-based stimulus control, to improve communication efficiency while reducing the number of user selections required. ChatBCI has the potential to significantly enhance the speed, accuracy, and usability of BCI-enabled communication systems for individuals with ALS and other neurological conditions affecting speech and motor function.
Market Applications:
- Assistive communication tools for ALS and other neurodegenerative disease patients.
- Rehabilitation technology and neuroprosthetics.
- Healthcare settings for patient-caregiver communication support.
- Adaptive technologies for individuals with severe motor impairments.
- Integration in AAC software and brain-computer interface device markets.
Advantages:
- Integrates large language models for semantically rich and contextually accurate communication.
- Utilizes EEG-based P300 signals for precise character selection without physical input.
- Employs word prediction and completion to accelerate typing speed.
- Adaptive stopping strategy minimizes stimulus duration, improving efficiency.
- Reduces cognitive effort, making communication easier for users with severe disabilities.
Publications:
- Hong, Najafizadeh et al.,2024. ChatBCI, a P300 speller BCI leveraging large language models for improved sentence composition in realistic scenarios
- Hong, Najafizadeh et al.,2025. ChatBCI-4-ALS: A High-Performance, LLM-Driven, Intent-Based BCI Communication System for Individuals with ALS
- Hong, Najafizadeh et al.,2026. ChatBCI, a P300 speller BCI with context-driven word prediction leveraging large language models, from concept to evaluation.
Intellectual Property & Development Status: Provisional application filed. Patent pending. Available for licensing and/or research collaboration. For any business development and other collaborative partnerships, contact: marketingbd@research.rutgers.edu
