A team led by University of Massachusetts Amherst researchers has developed a wearable wrist device powered by a machine-learning algorithm that can continually track changes in arm movement impairment caused by strokes. Monitoring changes throughout the entire rehabilitation process will allow clinicians to make real-time adjustments to therapy programs for tailored interventions and personalized care instead of the one-size-fits-all current standard. 

Image

“We are the first group to actually show that, using wearable data, we can extract information about patients’ motor severity, which clinicians can actually use to determine whether their intervention is effective or not,” says Sunghoon Ivan Lee, associate professor in the Manning College of Information and Computer Sciences at UMass Amherst and corresponding author on the paper describing this new technology. The research was completed alongside colleagues from Washington University in St. Louis, Shirley Ryan AbilityLab and Harvard Medical School/Mass General Brigham.

Annually, more than 795,000 Americans experience strokes, with upper-limb mobility issues affecting up to 77% of patients immediately following the event. About 40% of patients continue to have chronic issues, posing a major limitation to independent living.

While physical therapy improves these mobility issues, it is not without its limitations. Currently, progress is measured by a clinician’s observational assessment, which takes about 30 minutes. Because this is a time-consuming process, assessment typically occurs only pre- and post-rehabilitation.

“That means during that therapy process, neither the patient nor the therapist has a clear idea of how patients are responding to the treatments that they’re receiving,” says Lee. “Currently, clinicians aren’t able to see if patients are responding to the prescribed exercises, and patients have no way of knowing how they are progressing.”

With a wearable monitor, recovery data collection can be ongoing and occur outside of therapy sessions, allowing the therapist to make timely adjustments to treatment strategies for truly personalized rehabilitation.

Also, tracking movement in a patient’s real-life environment may be more indicative of true performance, as opposed to movement that is artificially produced in a clinic. The wrist-wearable also captures movement at all times of day, versus at just one snapshot of time. And finally, patients may find that tracking their own recovery progress increases their engagement in the therapy practices and keeps them motivated.

Lee is optimistic that this increased transparency will translate to improved therapy outcomes.

An accelerometer sensor in the device captures upper-limb movement, which is then interpreted by a machine-learning algorithm developed by Lee and his graduate student and the lead author on the paper, Ryan Wang.

“[Movement and impairment severity] are related because the less severe you are, the more likely you’re going to move a lot, but they’re not exactly the same,” says Lee. “Increasing the use of the limbs—yes, we can encourage the person to make use of the limb more. But patients cannot make instant changes to motor severity through short-term behavior change.”

Their model, described in Science Translational Medicine, was trained on accelerometer data and clinician assessment scores of subacute (one week to six months after a stroke) stroke patients and healthy individuals. They found that their algorithm was 40-50% more accurate—meaning a better reflection of the patient’s true condition—compared to clinician evaluation.

In addition to the clinical application of using the device to inform personalized interventions, Lee’s work demonstrated that the device has a strong research application. In recreating a previous study using their own digital biomarker instead of clinician observations, the researchers found that they generated statistically significant results with 50% fewer participants.

“We can get a clear idea of the effectiveness of the intervention using a lower sample size and far fewer resources,” says Lee. This can expedite the research process and reduce costs.”

This work was supported by the National Institutes of Health. With a provisional patent filed, Lee is pursuing commercialization through the startup Lumid Health with support from UMass Amherst Institute for Applied Life Sciences’ Translational Seed Award, the UMass Office of Technology Commercialization & Ventures’ (OTCV) Technology Development Fund and participation in the NSF I-Corps Training Program.

To further develop the technology, Lee’s research partners are currently recruiting stroke patients for a study held at the Spaulding Rehabilitation Hospital in Boston. Those interested in participating can see the inclusion criteria and apply here.