Body Movement Monitoring for Parkinson’s Disease Patients Using A Smart Sensor Based Non-Invasive Technique

Abstract

There have been increasing interests in recent years on using smart sensor technology, e.g., Kinect and Leap Motion, to capture and analyze human body movements, with the goal to benefit not only games, but also health care and rehab applications. We propose a non-invasive approach using movement data captured from Kinect to monitor motor deficits of Parkinson’s disease (PD) patients. We captured and evaluated simple exercises, normally performed in rehabilitation sessions by physical therapist: Stride Length, Tremor and Timed Up & Go (TUG). The standard medical UPDRS scale is used by a physical therapist to determine the level of severity as the ground truth. The general framework after getting the motion data includes two steps feature extraction from the kinematic motion data, and classification using random forest (RF) (for the stride length and tremor data) and K-means (for the TUG data). Our technique was validated by inviting a group of subjects whose kinematic data are used for PD motion analysis. The experimental results demonstrate the high accuracy of our approach in the assessment of PD using kinematic motion data. Our technique is also suitable in a remote monitoring environment, where data collected can be transmitted to experts for assessment.