New Robotic System Assesses Mobility After Stroke

Worldwide, strokes affect over 15 million individuals annually, leaving three-quarters of survivors with arm and hand limitations, including weakness and paralysis. Overcoming the tendency to underuse the affected arm, a phenomenon known as "arm nonuse" or "learned nonuse," is crucial for rehabilitation, but gauging arm usage outside clinical environments poses a significant challenge. Observing natural behavior often requires discreet monitoring methods. Addressing this need, researchers have now designed an innovative robotic system that collects accurate data on how stroke survivors spontaneously use their arms.

Developed by a team at USC Viterbi in Los Angeles, CA, USA, this cutting-edge approach employs a robotic arm to gather 3D spatial data about arm movements. The system utilizes machine learning algorithms to analyze this data, producing a reliable "arm nonuse" metric that can greatly assist clinicians in assessing rehabilitation progress. To make the experience engaging and supportive, a socially assistive robot (SAR) offers instructions and encouragement throughout the process. In their study, the USC Viterbi team worked with 14 participants who had been right-hand dominant prior to experiencing a stroke. The participants began by placing their hands on a 3D-printed box equipped with touch sensors, which served as the system's starting position. The SAR introduced the system's functionality and provided positive feedback. The robot arm would then move a button to various predetermined locations, initiating the "reaching trial" when the button lit up and the participant was cued to move.

Image: The robotic arm provides precise 3D spatial information to help stroke survivors (Photo courtesy of USC Viterbi)

The trial consisted of two phases: first, participants used their naturally preferred hand, mimicking typical daily activities. In the second phase, they were instructed to use their stroke-affected arm, akin to exercises performed in therapy or clinical settings. The team's machine learning analysis focused on three key metrics: the probability of arm use, the time taken to reach the target, and the successful completion of the reach. The study revealed significant differences in hand preference and time taken to reach targets among chronic stroke survivors. The method proved reliable over multiple sessions, with participants finding it easy to use and scoring it highly in terms of user experience.

Additionally, all participants deemed the interaction safe. The team received feedback suggesting that future enhancements could include personalized features, integrating additional behavioral data, and varying the tasks. This innovative approach not only demonstrated consistency and positive user experiences but also highlighted variations in arm use among participants. These insights are vital for healthcare professionals to more accurately monitor and facilitate stroke recovery.

“This work brings together quantitative user-performance data collected using a robot arm, while also motivating the user to provide a representative performance thanks to a socially assistive robot,” said Maja Matarić, study co-author and Chan Soon-Shiong Chair and Distinguished Professor of Computer Science, Neuroscience, and Pediatrics. “This novel combination can serve as a more accurate and more motivating process for stroke patient assessment.”

Related Links:
USC Viterbi