Home Monitoring of Sleep Disturbances in Parkinson's Disease: A Wearable Solution

Irene Rechichi, Luca Di Gangi, Maurizio Zibetti, Gabriella Olmo

Abstract: Sleep Disorders are the most common and disabling non-motor manifestations of Parkinson’s Disease (PD), significantly impairing the quality of life. Monitoring sleep disturbances in PD is a complex task, given the lack of objective metrics and the infrequent neurological assessments. This study proposes a framework for the detection of PD sleep patterns from data collected from 40 subjects (12 PD) through a wearable inertial measurement unit (IMU) during sleep, as well as the automatic assessment of sleep quality. Several features describing overnight motility are proposed and employed in Machine Learning (ML) models to carry out the classification. The best model achieved a 96.2% Accuracy and 93.4% F-1 score in detecting PD subjects from controls, in a Leave-One-Subject-Out cross-validation approach. Sleep quality was assessed with an average accuracy of 79.7% ± 4.4 across the three tested classifiers, and 75% ± 5.25 F-1 score. This suggests the feasibility of characterising overnight motility in PD and effectively monitoring the symptoms’ progression through lightweight technology, in a pervasive, e-Health scenario.