Visualization and Qualitative Analysis of Rehabilitation Exercises Based on a Mobile App

An average of 795,000 people have a stroke each year in the US and stroke is the leading cause of disability. Even though repetitive rehabilitation exercises on a regular basis are the key for motor recovery after stroke, only about one-third of patients follow prescribed exercise regimens. With the development of computer technologies and wearable devices, numerous mobile-based rehabilitation solutions have been proposed. However, how to quantify observations of motor recovery during and after exercise remains a challenge. Typically, if the smoothness of a patient's movement is improved, it indicates the patient's motor recovery. But the equipment to test the movement smoothness can be costly and inefficient, therefore not suitable for the home environment. In this paper, we introduce a mobile based rehab solution to support home exercises for stroke survivors and demonstrate its effectiveness by visualizing and qualitatively analyzing movement data captured by the proposed app. Specifically, we compute smoothness of movement for 15 patients over a 6-week user study. Exercises are designed and monitored by Occupational Therapy domain experts. We compute smoothness using log-dimensionless jerk (LDLJ) and Spectral Arc Length (SPARC) methods. Additionally, we study the Range of Motion (ROM) for each movement and explore the correlation between ROM and upper extremity movement.