Multidimensional Circadian Monitoring by Wearable Biosensors in Parkinson's Disease

Front Neurol. 2018 Mar 26:9:157. doi: 10.3389/fneur.2018.00157. eCollection 2018.

Abstract

Parkinson's disease (PD) is associated with several non-motor symptoms that may precede the diagnosis and constitute a major source of frailty in this population. The digital era in health care has open up new prospects to move forward from the qualitative and subjective scoring for PD with the use of new wearable biosensors that enable frequent quantitative, reliable, repeatable, and multidimensional measurements to be made with minimal discomfort and inconvenience for patients. A cross-sectional study was conducted to test a wrist-worn device combined with machine-learning processing to detect circadian rhythms of sleep, motor, and autonomic disruption, which can be suitable for the objective and non-invasive evaluation of PD patients. Wrist skin temperature, motor acceleration, time in movement, hand position, light exposure, and sleep rhythms were continuously measured in 12 PD patients and 12 age-matched healthy controls for seven consecutive days using an ambulatory circadian monitoring device (ACM). Our study demonstrates that a multichannel ACM device collects reliable and complementary information from motor (acceleration and time in movement) and common non-motor (sleep and skin temperature rhythms) features frequently disrupted in PD. Acceleration during the daytime (as indicative of motor impairment), time in movement during sleep (representative of fragmented sleep) and their ratio (A/T) are the best indexes to objectively characterize the most common symptoms of PD, allowing for a reliable and easy scoring method to evaluate patients. Chronodisruption score, measured by the integrative algorithm known as the circadian function index is directly linked to a low A/T score. Our work attempts to implement innovative technologies based on wearable, multisensor, objective, and easy-to-use devices, to quantify PD circadian rhythms in huge populations over extended periods of time, while controlling at the same time exposure to exogenous circadian synchronizers.

Keywords: Parkinson’s disease; circadian rhythms; machine learning; non-motor symptoms; sleep; wearable; wrist temperature.