Stanford researchers: Sustained engagement is a challenge for ResearchKit app

By Jonah Comstock

MyHeart Counts, one of the original five Apple ResearchKit studies, has had its first publication, in JAMA Cardiology. The paper, published yesterday, is mainly a feasibility study for large scale smartphone-based data collection, but also found some interesting correlations between physical activity and heart health. Ultimately, they collected data from more than 40,000 people between March and October 2015.

Researchers from Stanford University report pros and cons to a smartphone study. The advantages include the ability to rapidly collect a huge volume of data -- in a few weeks, the study was able to generate a 6-minute walk test data set larger than any ever collected -- and, for the most part, to be able to collect continuous fitness data instead of relying on periodic self-reports. Of course, researchers noted, not every participant keeps their smartphones on them at all times, so the continuous activity data isn't perfect.

On the other hand, researchers noted there was a pretty significant drop in engagement for many participants. 

"[T]he bar for entry to this study was much lower than that for equivalent studies performed using in-person visits," researchers wrote. "This lowering has the demonstrated advantage that many people consented but has the notable disadvantage that those individuals are by definition less invested in the study and thus less likely to complete all portions. For some data points in this study, we have data for only several thousand individuals, while almost 50,000 consented. We believe that the low bar in fact represents an opportunity to engage this larger group who are interested enough to download the app and answer a few questions but not much more. Balancing engagement, data feedback, and study design remain areas for further research."

The researchers do provide a few ideas for improving engagement, which include both EHR integration and Pokémon Go inspiration.

"We delayed the 6-minute walk test and heart age assessment until completion of all other portions of the study to minimize bias from this information, but that certainly contributed to the drop in participation in these tasks. An easy method to link lipid values directly from one’s electronic health record would help. However, even in the Practice Innovation and Clinical Excellence (PINNACLE) electronic health record–based cardiovascular registry, data to calculate the 10-year risk score were available in less than 30 percent of patients. Future versions of MyHeart Counts will introduce more personalization and earlier participant feedback. Elements of gamification, exemplified by Pokémon Go, could also be introduced to maximize engagement."

The sample was also not representative of the population. It was overwhelmingly (82 percent) male and also skewed young (average age was 36). Although participants came from all 50 states, nearly 10 percent came from California. 

Beyond the feasibility aspect, the study also contained some insights based on the data about heart health and physical activity that it collected. Maybe the most interesting was that the data shows heart health correlates not just to absolute activity but to frequent state changes from activity to nonactivity. Additionally, people were very likely to either underestimate or overestimate their heart health risks, but not to predict it accurately.