Related Experiment Video
Updated: Aug 22, 2025
11:54
Real-Time Proxy-Control of Re-Parameterized Peripheral Signals using a Close-Loop Interface
Published on: May 8, 2021
4.6K
Real-time brain-machine interface in non-human primates achieves high-velocity prosthetic finger movements using a
Matthew S Willsey1,2, Samuel R Nason-Tomaszewski2, Scott R Ensel2
1Department of Neurosurgery, University of Michigan, Ann Arbor, MI, USA.
Nature Communications
|November 13, 2022
Summary
Researchers developed a novel neural network algorithm to improve brain-machine interfaces for prosthetic limbs. This new algorithm significantly enhances the speed and naturalness of prosthetic finger movements, offering better motor function restoration.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Brain-machine interfaces (BMIs) aim to restore motor function but prosthetic performance lags behind native capabilities.
- Algorithmic limitations in translating brain signals to prosthetic control hinder realistic limb and finger movements.
Purpose of the Study:
- To develop and evaluate a shallow feed-forward neural network for decoding real-time finger movements.
- To improve the speed, naturalness, and overall performance of brain-controlled prosthetics.
Main Methods:
- A shallow feed-forward neural network was designed to decode two-degree-of-freedom finger movements.
- A two-step training method, including recalibrated feedback intention-training (ReFIT), was employed.
- Performance was tested across two adult male rhesus macaques over 7 days.
Main Results:
- The neural network decoders achieved a 36% increase in throughput compared to the standard ReFIT Kalman filter.
- Decoders demonstrated higher-velocity and more natural-appearing finger movements.
- Real-time decoding of continuous movements surpassed current state-of-the-art performance.
Conclusions:
- The developed neural network decoders represent a significant advancement in BMI technology for motor function restoration.
- This approach offers a promising foundation for creating more naturalistic and effective brain-controlled prosthetic devices.
