Objective: mPower is a clinical observational study of Parkinson’s disease (PD) conducted completely through an iPhone app focused on the ability of mobile devices and sensors to measure symptoms of PD and detect variations within those symptoms.

Background: For nearly a century, there have been engineering efforts to make quantitative measures of symptoms of PD. Despite many advances, these measures have provided minimal clinical impact. We hypothesize that quantitative measures may have the greatest impact when providing frequent repeated measures of symptoms over time. mPower is a smartphone app made freely available on iPhone for users to donate information through surveys and frequent sensor-based recordings from participants with and without PD.

Methods: All iPhone users in the US over 18 years of age were eligible to participate by downloading the mPower app and providing informed consent. All activities within the app are considered optional. In addition to a background survey, participants had four separate structured activities which leverage the sensors within the iPhone to make recordings. These activities (and the sensors) include a speeded tapping (iPhone screen and accelerometer), gait (accelerometer), balance (accelerometer), turning (accelerometer), phonation (microphone) activity. Features were extracted from each recording, and random forest was used for classification.

Results: mPower was launched in the US app store in March 2015, with an enrollment of over 15,000 participants in the first nine months. Of those who responded to the background survey, over 1700 participants indicate a diagnosis of PD with over 9000 controls enrolling. Despite attrition, about 25% of users maintain at least weekly use of the app over those nine months. Features derived from the performance of age-matched patients and controls on each of four activities was used to classify those with PD (tapping AUC 0.85, gait AUC 0.88, turning and balance AUC 0.71, phonation AUC 0.84). Furthermore, these same features personalized differences in response to medications in a subset of PD patients who provided the most data.

Conclusions: Frequent assessment of symptoms of PD can be made with personal smartphone devices using app-based studies. The features extracted from activities included in mPower provide a personalized quantified measure of phenotypic changes in PD that can be clinically validated.

« Back to 2016 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/mpower-a-smartphone-based-study-of-parkinsons-disease-provides-personalized-measures-of-disease-impact/