A High-Performance Neural Prosthesis Incorporating Discrete State Selection With Hidden Markov Models

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Communication neural prostheses decode neural activity to restore communication for individuals with motor neurological injuries.
• Current high-performing prostheses primarily rely on analog decoding, often lacking robust discrete state decoding.
• Seamless integration of both analog and discrete decoding is crucial for intuitive and high-performance prostheses.

Purpose of the Study:

• To incorporate a hidden Markov model (HMM) into an intracortical communication prosthesis.
• To enable accurate and fast discrete state decoding in parallel with analog decoding.
• To enhance the performance of communication neural prostheses.

Main Methods:

• Implemented a hidden Markov model (HMM) within an intracortical neural prosthesis system.
• Conducted closed-loop experiments with nonhuman primates using multielectrode arrays.
• Evaluated the impact of the HMM's transition model on decoding performance.

Main Results:

• Incorporating the HMM increased state-of-the-art achieved bitrate by 13.9% and 4.2% in two monkeys.
• The HMM's transition model was critical for performance enhancement.
• Achieved peak bitrates of 6.5, 5.7, and 4.7 bits per second in Monkeys J, R, and L, respectively.
• Demonstrated robust controllability of the neural prosthesis throughout experimental sessions.

Conclusions:

• High-performance discrete decoding can be beneficially combined with analog decoding in neural prostheses.
• The integration of HMMs represents a significant advancement in neural prosthesis technology.
• This approach achieves new state-of-the-art performance levels for communication restoration.