New ‘Intelligent Suit’ Should Greatly Improve Spinal Cord Injury Treatment
A new rehabilitation suit driven by artificial intelligence (AI) should provide much more effective treatment after a serious spinal cord injury.
Like existing SCI rehab suits, the new suit will combine functional electrical stimulation (FES) with mobility support using artificial tendons. But unlike existing suits, this one has a superpower: It will use AI to react to patients’ intended movements.
The suit will be developed over the next several years by researchers from Friedrich Alexander University (FAU) in Germany, working with Heidelberg University and Heidelberg University Hospital. The German Research Foundation has provided about one million Euros to fund the project, including 357,000 Euros for FAU.
Standard SCI Treatments Aren’t Good Enough
FAU points out that incomplete spinal cord injuries can be partially reversed if the long nerve fibers aren’t separated entirely from the brain, thus leaving some remnants of motor control.
“In such cases, there is a good chance of regaining at least some mobility,” FAU professor Dr. Claudio Castellini says in a statement. “The potential for regeneration is particularly high in the first twelve months after the injury, when new neurons can be generated and new [neural] networks created.”
That said, conventional treatments for a spinal cord injuries leave a lot to be desired.
Standard SCI treatments are mainly based on repeated mobility exercises that require patients to grasp and move an object, for example. Therapists also use FES, which causes specific muscles to contract.
Another approach is to fit a patient with an exoskeleton or suit. Typically, these are orthoses that support movement using pulleys or inflatable air chambers.
These treatments are often better than nothing and can help a patient to partially recover. However, Castellini says that although great progress has been made in recent years, current therapies still don’t track as well as they could with how the brain actually learns to move various parts of the body.
This is true primarily for three reasons: Especially weak muscles often aren’t targeted properly, patients aren’t given motor exercises that are actually useful, and rehab isn’t tailored to patients’ individual needs. Patients can lose interest in the treatment as a result.
AI-Driven Suit Recognizes Intended Movements
Over the next few years, Castellini and others will seek to develop a new rehabilitation suit that should make post-SCI therapies much more effective.
The “exo-suit” they’re working on consists of a compression jacket, arm cuffs, and gloves, similar to other rehab suits. However, it has one key additional feature: built-in AI technology that recognizes the movements a patient is trying to make.
The exo-suit uses integrated sensors to measure muscle activity, says Marek Sierotowicz, a doctoral candidate involved in the project. Computer algorithms use these measurements to determine what the patient wants to do and adjust the rest of the suit accordingly.
In other words, the AI tells the FES system and the exo-suit where muscle contractions should be triggered or pulleys tightened in order to support the intended movement. This should result in therapies that are much better targeted to specific patients.
Training the AI
For the AI to work, it first has to be trained. To do that, researchers are building a complete virtual model of the muscles and skeleton of someone without a disability.
“We will carry out our initial tests with people who are not disabled and gather as much data as possible,” Sierotowicz says. “The better we train AI, the more reliably we will be able to recognize movement patterns and the more accurately the assistance systems will be able to work in the future.”
It’s also hoped that the AI will allow the EFS and robotic pulley systems to work better together, which should make for gentler treatments. Using FES alone requires intense simulation that patients often find unpleasant, the researchers noted.
As always, I’ll stay on top of this and other spinal cord injury news and report any new developments. In the meantime, what are the limitations you’ve found with existing SCI treatments? Do you think the AI approach is promising? Please share your views in the comments. TwP