Wake/Sleep Identification Based on Body Movement for Parkinson’s Disease Patients

Abstract

This paper presents a method for identifying a patient’s wake/sleep state for closed-loop deep brain stimulation (DBS). The method uses a real-time wake/sleep identification algorithm that includes posture analysis based on the movement of the chest below the clavicle, which is the location of the subcutaneous pulse generator. A single micro-accelerometer was used to monitor the movement of the wrist and the chest of thirteen healthy adults and twelve patients with Parkinson’s disease for nine continuous hours. The wake/sleep state identification for the chest algorithm had accuracy, sensitivity, and specificity values of 85.78, 84.21, and 82.08 %, respectively, compared to video recordings for patients with DBS ON, and 82.74, 82.68, and 82.28 %, respectively, for patients with DBS OFF. The algorithm performance for the chest is comparable to that of the commonly used location on the wrist. The real-time wake/sleep identification algorithms were proved to be effective. This research provides a practical method for closed-loop DBS, which will greatly benefit patients with Parkinson’s disease.