Skip to content

Is AI revolutionizing rehabilitation care? This Northeastern expert is digging deep on the issue   

Gene Tunik and Matthew Yarossi conducting research.
Northeastern professors Gene Tunik and Mathew Yarossi at work in the university’s Laboratory for Movement Neurosciences. Photo by Matthew Modoono/Northeastern University

A good physical therapist has several key, very human traits, explains Mathew Yarossi, a Northeastern University professor of electrical and computer engineering, and physical therapy, movement and the rehabilitation sciences.  

Physical therapists are attentive and observant to a patient’s needs, closely examining their range of motion and ability to complete an exercise or task. The job also requires adaptability and strong communication skills. If a patient is unable to complete a certain movement, for example — whether because of pain or another unrelated issue — the therapist must readjust and clearly communicate the new plan to the patient.

It’s a constant “task dynamics loop” in which therapists are observing and reacting, Yarossi says. 

“The therapist is continuously monitoring the patient for safety, for performance, fatigue, affect,” Yarossi says. “They’re providing feedback on performance. They’re motivating the patients. They’re educating the patient. The therapist is doing all these intelligent actions.” 

So how does artificial intelligence and robotic technologies enter the mix in a job that requires such innate and dynamic human qualities? 

These systems can play a critical role in providing therapists with key data to understand the patient’s neuromuscular function as they complete these exercises, enabling “a new brain “dynamic loop” to occur in sequence with the task dynamics loop. This can play a key role in the development of more targeted and effective treatment plans, Yarossi, who works in the university’s Laboratory for Movement Neurosciences, says.

Let’s say a patient moves an arm up and down as part of an exercise plan prescribed by a therapist. With the use of a VR headset or a robotic therapeutic device, the patient can be given direct feedback on the action without the need for a person on hand. While completing those tasks, the patient can be outfitted with sensors that allow the clinicians to detect how muscles are reacting to the movement, as well as the activity going on in the brain, Yarossi explains. 

“With this sensing technology, we can get the information that the therapist doesn’t have with their eyes and their hands,” he says. 

From there, the therapists can use brain stimulation devices to inhibit or activate electrical activity in the brain to assist the patient in their rehabilitation treatment plan. 

“Perhaps this is the thing we need to do to get stimulation, robotics and VR into the clinic,” Yarossi says. “We need to close loopholes.” 

Yet there are challenges in bringing these kinds of technologies to your local clinic, Yarossi explains. 

A therapist’s time with a patient is often limited to the duration of their session. If setting up these systems takes too long between patients, it’s basically dead on arrival, Yarossi says. Additionally, these systems have to be functional nearly 100% of the time to be used in a health care setting. 

The future depends on adapting these technologies and increasing their usability and functionality, he notes. But that will occur by inviting multiple stakeholders to the table, ranging from computer scientists and AI experts to doctors and physical therapists. 

“AI is going to change the way we work,” he says. “But how we do it is up to us, and collaboration is really important.”