Wearable sensors enable continuous monitoring in daily life, but how can they provide insights into the progression of Parkinson’s disease? In this project, researchers developed models to quantify rest tremor, reduced arm swing when walking and heart rate changes, based on raw sensor data from a smartwatch. These models were specifically designed to cope with real-life variability, for example by measuring arm swing only when no other arm activities were detected.
Using two years of continuous sensor data from the Personalized Parkinson Project, the researchers identified a set of digital biomarkers to capture the week-by-week progression in early-stage Parkinson’s disease. Examples include the portion of the day when tremor was present, the maximum arm swing amplitude, and the peak heart rate during the day.
These digital biomarkers show promise to be used in clinical trials investigating treatments to slow down disease progression. The hope is that such innovative outcome measures will accelerate the discovery of new therapies for people living with Parkinson’s disease. The models are available through an open-source Python toolbox called ParaDigMa (pypi.org/project/paradigma/), enabling researchers worldwide to apply them to their own study data.