Providing Real-Time Exercise Feedback to Patients Undergoing Physical Therapy

Abstract

Musculoskeletal conditions, often requiring rehabilitation, affect one-third of the U.S. population annually. RehabBuddy is a rehabilitation assistance system that extends the reach of a physical rehabilitation specialist beyond the clinic. This thesis presents a system that uses body-worn motion sensors and a mobile application that provides the patient with assistance to ensure that home exercises are performed with the same precision as under clinical supervision. Assisted by a specialist in the clinic, the wearable sensors and user interface developed allow the capture of individualized exercises unique to the patient's physical abilities. Beyond the clinical setting, the system can assist patients by providing real-time corrective feedback to repeat these exercises through a correct and complete arc of motion for the prescribed number of repetitions. An inertial measurement unit (IMU) is used on the body part to be exercised to capture its pose. Presented is a kinematics data processing approach to defining custom exercises with flexibility in terms of where it is worn and the nature of the exercise, as well as real-time corrective feedback parameters. This thesis goes through the engineering approach, initial student investigator trials, and presents new preliminary subject data from subject trials currently ongoing at the University of Kentucky. The system is tested on multiple exercises performed by multiple subjects. It is then demonstrated how it can improve exercise adherence by assisting patients in reaching the full prescribed range of motion and avoid overextension, assist in adherence to the ideal plane of motion, and affect hold time.