Wearable Sensors May Help Predict On, Off States in Parkinson Disease

parkinson's disease
parkinson’s disease
Wearable technology may help monitor on/off states in patients with Parkinson disease in an outpatient setting.
The following article is part of conference coverage from the 2018 American Academy of Neurology Annual Meeting in Los Angeles, California. Neurology Advisor’s staff will be reporting breaking news associated with research conducted by leading experts in neurology. Check back for the latest news from AAN 2018.

LOS ANGELES – Wearable sensor technology may be promising for detecting on and off states in patients with Parkinson disease, according to data presented at the 2018 American Academy of Neurology Annual Meeting, April 21-27 in Los Angeles.

Investigators evaluated 10-meter Instrumented Stand and Walk (ISAW) tests during on and off periods in 25 patients with Parkinson disease who were taking levodopa. The patients also wore Ambulatory Parkinson Disease Monitory (APDM) sensors on their sternum, wrists, lumbar, and lower extremities. Each ISAW was scored by a neurologist based on the Unified Parkinson Disease Rating Scale-3 (UPDRS-3). Ultimately, 98 kinematic features were analyzed for significance to the neurologist-evaluated total motor score and patient-reported on/off periods via statistical models and machine learning methods.

The investigators found that 22 kinematic features were significantly different compared with patient-reported on/off periods. The most significant were trunk transverse range of motion, arm range of motion, mid-swing elevation, stride length, turn velocity, steps in turn, and toe out angles.

Regression models showed an average difference of 14 points between on/off periods in total UPDRS-3, and 9 points when adjusting for 5 significant features from individual baselines, including mean trunk transverse range of motion, right arm range of motion, and toe out angle. Based on these features, the investigators used the following approaches to predict on/off periods: direct binary classification (acc=0.56), regression to total UPDRS score (acc=0.76), regression to Postural Instability and Gait Disorder sub-score (acc=0.64), and classification of on-off/off-on transitions using feature differences (Naïve Bayes: acc=0.74, area under the curve (AUC) =0.78; Random Forest: acc=0.76, AUC=0.90).

Related Articles

While the results will need to be confirmed in larger studies, the investigators concluded that this technology “could be particularly useful for monitoring response to therapy in an outpatient setting.”

Disclosures: Several authors report professional relationships and employment with IBM T.J. Watson Research Center and Pfizer Inc.

For more coverage of AAN 2018, click here.


Anand V, Bilal E, Ramos V, et al. Automatic detection of on/off states in Parkinson disease patients using wearable sensor technology. Presented at: 2018 American Academy of Neurology Annual Meeting. April 21-27, 2018; Los Angeles, CA. S3.007.