Detecting Patient Engagement in Robotic Rehabilitation Using IMU and sEMG Signals

Authors: Sabrina Johnson, Ehsan Esfahani

SUNY Campus: SUNY Buffalo

Presentation Type: Poster

Location: Old Union Hall

Presentation #: 23

Timeslot: Session D 3:00-4:00 PM

Abstract: In the United States, 795,000 people suffer from stroke annually, with approximately 80% sustaining motor impairments. Traditional physiotherapy has drawbacks such as intensity and uncertainty, leading to inadequate patient participation and inconsistent progress evaluation. Rehabilitation robotics has proven more effective in patient recovery. However, interfaces between users and robots are not intuitive due to the nature of human motion. We aim to use a classifier that detects patient engagement in physical therapy activities using IMU and sEMG signals. This information will be used to adjust the force exerted by a rehabilitation robot. The robot should provide more force when the patient is engaged but unable to move the robot's end-effector, and less force when the patient's engagement is passive or their force is adequate. The force parameter can be tuned in real time using the developed classifier, optimizing the patient's effort during therapy tasks and accelerating recovery time. This research is a starting point for machine learning in rehabilitation robotics using IMU and sEMG signals. The developed algorithm could serve as a foundation for more complex and intuitive computer-human interaction detection.