AI and Brain-Computer Interface May Predict Intended Motion in Those With Paralysis

Source: Geralt/Pixabay

Artificial intelligence (AI) machine learning combined with brain–computer interface (BCI) neurotechnology is offering a glimmer of hope for the disabled. A new study published in the Journal of Neurophysiology shows how an AI-enabled BCI has the potential to enable those with severe paralysis from spinal cord injury to control external devices by predicting intended movement using a wearable sensor.

One of the most severe forms of paralysis is quadriplegia, also known as tetraplegia, a condition in which there is partial or complete loss of movement in both arms and both legs caused by damage to the brain or due to spinal cord injury.

“This wearable system has the potential to enable people with tetraplegia to control assistive devices through movement intent,” wrote the researchers affiliated with Carnegie Mellon University, the University of Pittsburgh, Imperial College in London, and the Battelle Memorial Institute in Ohio.

No Surgery Required

Currently, intracortical BCIs, also known as brain–machine interfaces, require surgery, which is a barrier to widespread adoption. What makes this study unique is that the scientists have shown how a wearable sensor array combined with AI machine learning can provide a potential alternative to surgically implanted solutions.

“The objective of this study was to evaluate the feasibility of deriving motor control signals from a wearable sensor that can detect residual motor unit activity in paralyzed muscles after chronic cervical spinal cord injury (SCI),” the scientists wrote.

Myoelectric Activity

The scientists created a sleeve electrode array with 150 stainless steel disc electrodes on a stretchable fabric to measure myoelectric activity—the electrical activity generated by muscles. This sleeve was placed over the forearm of a 32-year-old quadriplegic man with spinal cord injuries from 14 years prior to the study. The study participant could not move his fingers but did have limited movement in his wrist.

“This study provides the first demonstration of a wearable interface for recording and decoding firing rates of motor units below the level of injury in a person with motor complete SCI,” wrote the researchers.

The researchers recorded the electrical pulses associated with muscle contractions generated while the quadriplegic man attempted to perform a series of gestures displayed on a computer screen as motor unit action potentials from 20 motor units were isolated. The movements attempted included the shoulder, fingers, wrist, and elbow in the first session. In a follow-on session 18 months after, the same participant was given video instruction to attempt movements of the fingers and wrist, hand-extension movements, and cylindrical grasp.

Even though the participant could not move his fingers due to his spinal cord injuries, there were muscles that were still connected to his brain that could generate myoelectric signals, albeit the signals were weak.

Linear Discriminant Analysis

The researchers used linear discriminant analysis, a commonly used pre-processing step for AI supervised machine learning that reduces dimensionality and classifies the data simultaneously.

“These results demonstrate the potential to create a wearable sensor for determining movement intentions from spared motor neurons, which may enable people with severe tetraplegia to control assistive devices such as computers, wheelchairs, and robotic manipulators,” concluded the scientists.

Copyright © 2022 Cami Rosso. All rights reserved.