Improving neural prosthetic system performance by combining plan and peri-movement activity.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Neural Engineering

Background:

• Current neural prosthetic systems often rely on limited neural activity data.
• Estimating arm movements typically uses either pre-movement (plan) or movement (peri-movement) activity.

Purpose of the Study:

• To investigate the impact of combining plan and peri-movement neural activity on decoding arm movements.
• To determine if incorporating time-varying structures in neural activity further enhances decoding accuracy.

Main Methods:

• Recorded neural activity from the pre-motor cortex of a rhesus monkey during a delayed-reach task.
• Utilized decoding classification algorithms to analyze neural signals from distinct time periods.
• Assessed the effect of combining plan and peri-movement activity, and incorporating time-varying structures.

Main Results:

• Decoding performance improved by 56% (plan activity) and 71% (peri-movement activity) when both were combined.
• Accounting for time-varying peri-movement activity improved performance by an additional 15%.
• Low correlations were observed between simultaneously recorded units and across time periods.

Conclusions:

• Combining neural information from both plan and peri-movement periods significantly boosts decoding performance for neural prosthetics.
• The independence assumption between units and time periods minimally impacts performance, simplifying decoding models.