Abstract
Background:
Wearable sensors have been used successfully to characterize bradykinetic gait in PD patients, but most studies have been conducted in highly controlled laboratory environments.
Objective:
To assess whether sensor-based analysis of real-life gait can be used to objectively and remotely monitor motor fluctuations in Parkinson’s disease (PD).
Methods:
The Parkinson@Home validation study provides a new reference dataset for the development of digital biomarkers to monitor persons with PD in daily life. Specifically, a group of 25 PD patients with motor fluctuations and 25 age-matched controls performed unscripted, typical daily activities in and around their homes for at least one hour, while being recorded on video. PD patients did this twice: after overnight withdrawal of dopaminergic medication; and again one hour after medication intake. Participants wore sensor devices on both wrists and ankles, on the lower back, and in the front trouser pocket, capturing both movement and contextual data. Gait segments of 25 seconds were extracted from accelerometer signals based on manual video annotations. The power spectral density of each segment and device was estimated using Welch’s method, from which the total power in the 0.5-10 Hz band, width of the dominant frequency and cadence were derived. The ability to discriminate between before and after medication intake, and between PD patients and controls, was evaluated using leave-one-subject out nested cross-validation.
Results:
For 18 PD patients (11 men, median age: 65 years, median MDS-UPDRS part III off: 42) and 24 controls (13 men, median age: 68 years), ≥10 gait segments were available. Using logistic LASSO regression, we could classify whether the unscripted gait segments occurred before or after medication intake, with mean AUCs varying between 0.70 (ankle of least affected side, 95% CI: 0.60 to 0.81) and 0.82 (ankle of most affected side, 95% CI: 0.72 to 0.92) across sensor locations. Combining all sensor locations did not significantly improve classification (AUC of 0.84, 95% CI: 0.75 to 0.93). Of all signal properties, the total power in the 0.5-10 Hz band was most responsive to dopaminergic medication. Discriminating between PD patients and controls was generally more difficult (AUC of all sensor locations combined: 0.77, 95% CI: 0.64 to 0.89). The video recordings revealed that the positioning of the hands during real-life gait had a substantial impact on the power spectral density of both the wrist and trouser pocket sensor.
Conclusions:
We present a new video-referenced dataset that includes unscripted activities in and around the participants’ homes. Using this dataset, we show the feasibility of using real-life gait to monitor motor fluctuations, based on a single sensor location. Future work may assess the value of contextual sensors to control for real-world confounders.
Wearable sensors have been used successfully to characterize bradykinetic gait in PD patients, but most studies have been conducted in highly controlled laboratory environments.
Objective:
To assess whether sensor-based analysis of real-life gait can be used to objectively and remotely monitor motor fluctuations in Parkinson’s disease (PD).
Methods:
The Parkinson@Home validation study provides a new reference dataset for the development of digital biomarkers to monitor persons with PD in daily life. Specifically, a group of 25 PD patients with motor fluctuations and 25 age-matched controls performed unscripted, typical daily activities in and around their homes for at least one hour, while being recorded on video. PD patients did this twice: after overnight withdrawal of dopaminergic medication; and again one hour after medication intake. Participants wore sensor devices on both wrists and ankles, on the lower back, and in the front trouser pocket, capturing both movement and contextual data. Gait segments of 25 seconds were extracted from accelerometer signals based on manual video annotations. The power spectral density of each segment and device was estimated using Welch’s method, from which the total power in the 0.5-10 Hz band, width of the dominant frequency and cadence were derived. The ability to discriminate between before and after medication intake, and between PD patients and controls, was evaluated using leave-one-subject out nested cross-validation.
Results:
For 18 PD patients (11 men, median age: 65 years, median MDS-UPDRS part III off: 42) and 24 controls (13 men, median age: 68 years), ≥10 gait segments were available. Using logistic LASSO regression, we could classify whether the unscripted gait segments occurred before or after medication intake, with mean AUCs varying between 0.70 (ankle of least affected side, 95% CI: 0.60 to 0.81) and 0.82 (ankle of most affected side, 95% CI: 0.72 to 0.92) across sensor locations. Combining all sensor locations did not significantly improve classification (AUC of 0.84, 95% CI: 0.75 to 0.93). Of all signal properties, the total power in the 0.5-10 Hz band was most responsive to dopaminergic medication. Discriminating between PD patients and controls was generally more difficult (AUC of all sensor locations combined: 0.77, 95% CI: 0.64 to 0.89). The video recordings revealed that the positioning of the hands during real-life gait had a substantial impact on the power spectral density of both the wrist and trouser pocket sensor.
Conclusions:
We present a new video-referenced dataset that includes unscripted activities in and around the participants’ homes. Using this dataset, we show the feasibility of using real-life gait to monitor motor fluctuations, based on a single sensor location. Future work may assess the value of contextual sensors to control for real-world confounders.
| Original language | English |
|---|---|
| DOIs | |
| Publication status | Published - 2 Apr 2020 |