IPhone-based Parkinson's study shares 6 months of data on 9,500 participants

smartphone up close
(Gonzalo Baeza / CC BY 2.0)

Sage Bionetworks has released the first tranche of data from its iPhone-based study of Parkinson's disease. With 9,520 participants consenting to share their results with researchers, the study dwarfs traditional trials on one level, but with fewer than 10% of people contributing data on 5 or more days, it trails conventional research in other ways.

When Apple ($AAPL) introduced its iPhone-based study framework, ResearchKit, last year, the first wave of trials quickly racked up enormous enrollment numbers. Nonprofit biomedical research team Sage was no exception. More than 12,000 people have signed up for its mPower Parkinson's study. Many traditional exploratory trials enroll fewer than 100 people. By using the sensors in iPhones to assess the dexterity, balance, memory and vocal characteristics of the 12,000 participants, Sage hopes to learn more about the effect of medication and other aspects of the disease.

The potential for such insights to be uncovered in the data is magnified by the willingness of people to share their results with all qualified researchers. More than 75% of participants have consented to such sharing, enabling Sage to release a 6-month trove of information on 9,520 people with the research community. "Our hope is that by sharing these data immediately, prior even to our own complete analysis, we will shorten the time to harnessing any utility that this study's data may hold to improve the condition of patients," Sage researchers wrote in Nature Biotechnology.

In enrolling a large number of people and then sharing their data widely, mPower is a poster child for the potential of ResearchKit. However, it is also an illustration of some of the questions hanging over the effectiveness of the model.

While 9,520 people consented to sharing data, more than 10% of these people never completed a single task or study. Around 70% of participants completed the enrollment study. Sage needed this data to see which participants self-identified as being diagnosed with Parkinson's. The dropping off of engagement with the study continued at each step of the process. Just shy of 900 people contributed data on 5 or more days. And, while some of these participants visited the app more than 150 times during the first 6 months of the study, most only checked in on 5 or 6 days.

The upshot is that Sage has a data set that is less colossal than the top-line enrollment figure implies,  but is still substantial compared to those generated in traditional studies. Sage has focused its initial analyses on the subset of relatively active users. "In a preliminary analysis for the first 6 months of data from those who used the app most, predictions of the effect of medication from these features reveal distinct patterns phenotypes," researchers wrote in an editorial published in NPJ Parkinson's Disease.

- read the release
- here's the Scientific Data paper
- check out the NPJ editorial
- and the Nature Biotechnology commentary

Suggested Articles

The new solution aims to streamline the incorporation of human genomic data into clinical trial designs.

TriNetX's platform uses EHR data to help drug developers with clinical trial protocol design and study site and participant identification.

The company's Owkin Socrates platform applies transfer and federated learning to clinical research.