Abstract P202: Estimation of Arm Kinematics in Stroke Survivors With Wearable Sensors

Introduction: Stroke is the leading cause of long-term disability in the US, often resulting in upper extremity (UE) motor impairment. Numerous clinical metrics have been developed and used to evaluate UE motor function to assess rehabilitation outcomes, but most of these are innately subjective and/or have relatively low sensitivity to change due to the use of ordinal scales. As a result, there is a growing interest in the use of kinematics to evaluate UE motor function, as they provide completely objective, high resolution quantitative measurements. However, obtaining kinematic measurements in stroke survivors can be a challenging task, as non-ideal environments (e.g., hospital rooms), expense, and a limited ability to perform calibration poses can make traditional optical tracking systems impractical. To overcome these challenges, we developed and compared two methods using different wearable sensors to estimate elbow angle and wrist position during reaching movements in people with stroke. Methods: We developed frameworks specific for two different types of sensors, inertial measurement units (IMU) and virtual reality (Vive) trackers, to estimate elbow angle and wrist position. We assessed each sensor’s estimation accuracy during pure flexion-extension motion, performed by a manual goniometer, and pure pronation-supination motion, performed by a healthy participant. We also compared each sensor’s ability to longitudinally track a stroke survivor’s UE function by using the wrist position estimates to compute sweep areas during a sweeping task. Results: For the IMUs, our results demonstrated accuracy to within 4.8° and 1.2 cm for elbow angle and wrist position, respectively. For the Vive, our results demonstrated accuracy to within 2.2° and 0.9 cm. The change in sweep area estimated by each method agreed with each other, as the difference between estimates was approximately 4% of the change in sweep area. Conclusion: These methods can be used to develop and accurately assess kinematic metrics for use in evaluating stroke rehabilitation outcomes.