Quantification of Compensatory Torso Motion in Post-Stroke Patients Using Wearable Inertial Measurement Units

Occupational Therapy (OT) tasks performed at home are not usually observable by a clinician. For remote environments, wearable sensor technologies could provide quantitative assessment of movement strategies to support clinical evaluation via telemedicine. This work presents an algorithm for estimating torso orientation and an individualized metric of compensatory torso motion for upper extremity tasks using an Inertial Measurement Unit (IMU) worn on the sternum. Two post-stroke males with hemiparesis and two healthy age-matched males were evaluated during a series of OT assessments, including active range of motion, pinch strength, standing balance, Moberg pick-up test, Nine-hole peg test, and a custom peg board grasp task. Torso orientation was estimated by decomposing IMU orientation into angles (pitch, roll, yaw) in each anatomical plane (sagittal, coronal, transverse) and referenced against standing balance posture. For each participant, a threshold of nominal torso motion was created from the variance in orientation for tasks performed with the Dominant or Unaffected arm. A percentage of time out of range (PTOR) of nominal variation was developed to distinguish operationally relevant differences. In this study, large PTOR values (greater than 30%) were consistent with OT observations of notable compensatory movement. IMU sensitivity to sensor-to-torso alignment was also evaluated in simulation and demonstrated interactions between pitch and roll angles when an IMU misalignment existed about the yaw axis. These case studies support the use of wearable IMUs to quantitatively assess compensatory torso motions and convey operationally relevant information about movement strategy to a clinician without the use of visual observation.