Parkinson\u27s Symptoms quantification using wearable sensors

Parkinson’s disease (PD) is a common neurodegenerative disorder affecting more than one million people in the United States and seven million people worldwide. Motor symptoms such as tremor, slowness of movements, rigidity, postural instability, and gait impairment are commonly observed in PD patients. Currently, Parkinsonian symptoms are usually assessed in clinical settings, where a patient has to complete some predefined motor tasks. Then a physician assigns a score based on the United Parkinson’s Disease Rating Scale (UPDRS) after observing the motor task. However, this procedure suffers from inter subject variability. Also, patients tend to show fewer symptoms during clinical visit, which leads to false assumption of the disease severity. The objective of this study is to overcome this limitations by building a system using Inertial Measurement Unit (IMU) that can be used at clinics and in home to collect PD symptoms data and build algorithms that can quantify PD symptoms more effectively. Data was acquired from patients seen at movement disorders Clinic at Sanford Health in Fargo, ND. Subjects wore Physilog IMUs and performed tasks for tremor, bradykinesia and gait according to the protocol approved by Sanford IRB. The data was analyzed using modified algorithm that was initially developed using data from normal subjects emulating PD symptoms. For tremor measurement, the study showed that sensor signals collected from the index finger more accurately predict tremor severity compared to signals from a sensor placed on the wrist. For finger tapping, a task measuring bradykinesia, the algorithm could predict with more than 80% accuracy when a set of features were selected to train the prediction model. Regarding gait, three different analysis were done to find the effective parameters indicative of severity of PD. Gait speed measurement algorithm was first developed using treadmill as a reference. Then, it was shown that the features selected could predict PD gait with 85.5% accuracy