Objective: Herein we present the results of ongoing work focused on estimating limb-specific dyskinesia in patients with PD using wearable sensor data collected in the clinic.

Background: A major challenge in the management of Parkinson’s diseases (PD) is to accurately monitor the severity of PD symptoms over time. The development of wearable sensors has opened the door for long-term monitoring of patients in the home and community settings. There is a need for developing methods suitable to monitor the severity of PD symptoms in patients with PD in and outside the clinic.

Methods: The motor function of ten patients diagnosed with idiopathic PD was assessed clinically on a 0-4 scale and using wearable sensors to collect acceleration data during the performance of upper-limb tasks. The tasks were repeated 24 times at 30 minute intervals over 2 days. Predefined features were extracted from the acceleration data of the limbs not performing voluntary movements. Then, a feature selection algorithm was used to identify features relevant to the estimation of the clinical scores. Finally, clinical scores of dyskinesia were estimated using a cost-sensitive random forest algorithm. A leave-one-subject-out cross-validation technique was used to estimate the clinical scores.

Results: From the acceleration data, we were able to estimate limb-specific scores of dyskinesia. Mean absolute deviation for the left arm, right arm, left leg, and right leg scores were 0.40, 0.29, 0.38, and 0.39, respectively.

Conclusions: The preliminary results of this study demonstrate the feasibility of estimating the severity of dyskinesia using wearable sensor data. While additional work is required to improve the accuracy of the estimation procedures, this work highlights the possibility of remotely assessing the severity of PD symptoms in patients with PD.

« Back to 2016 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/monitoring-dyskinesia-severity-using-wearable-sensor-data/