Applying Machine Learning To Treat Paralysis

Machine learning restores control for paralysis.


When Ian Burkhart was 19 years old, he was in an accident that damaged his spinal cord. He was paralysed from the chest down, losing all feeling in his hands and feet.

Screen Shot 2016-04-20 at 1.49.47 pmNow a team of scientists and doctors in the USA have been able to restore some control in Burkhart’s right hand through the use of innovative technology.

For five years Burkhart – who broke his neck when he dove under a wave at a beach in North Carolina – has been unable to do mundane tasks like stir a straw.

The procedure has enabled Burkhart, now 24, to use his hand and fingers, but it has not eliminated the paralysis.

Indeed Burkhart is entirely dependent on the technology providing the stimulus for any motor activity.

He told the New York Times that once his post-surgery rehabilitation commenced, he had to watch his hand to understand whether he was squeezing or wiggling his fingers since he had no sensation in his limb or digits.

The first major stage in Burkhart’s case began two years ago when doctors at the Ohio State Center for Neuromodulation planted a computer chip in his brain.

In the team’s report published in the science journal Nature the authors write that they applied machine learning algorithms to decode nerve cell activity and control in Burkhart’s right forearm.

A custom-built high-resolution neuromuscular electrical stimulation system, “provided isolated finger movements and [Burkhart] achieved six different wrist and hand motions.”

The software was developed by scientists at the Battelle Memorial Institute in Columbus, Ohio, a not for profit organisation with interests in a wide range of research activity including agri-business, energy, the environment, and national security.

With the kind of injury that Burkhart sustained, brain signals that provide movement to the body remain intact.

“But those signals as they arrive in the spinal cord are completely blocked… they can’t get to his muscles,” explains Dr Chad Bouton.

The lead author of the report, Bouton explains the science involved in a Nature demonstration video.

The chip in Burkhart’s brain provides a ‘window’ to his neural activity. Computer generated imagery offers a virtual ‘map’ to Burkhart’s thought patterns.

If he thinks ‘squeeze fingers’ the technology will display that. As Burkhart relearns how to use his motor skills, the computer’s machine learning software enables it to learn it too; it then feeds back this information to the device strapped to Burkhart’s wrist, which “drives hand and finger movements.”

The team involved have told media that the research is some way off to resolving the ultimate challenge for patients with paralysis: the restoration of completely independent movement.

Still, Burkhart’s story offers a breakthrough: “This is the first time a completely paralysed person has regained movement just be using their own thoughts,” said Bouton.

Researchers believe that the technology here can be applied to people who have suffered a stroke or brain trauma.

Still a lab-based project, the NY Times reports that funding for the project will cease in 2016.

Besides a cessation of research, this will mean that Burkhart will no longer have access to the technology.