Skip to main content
Premium Trial:

Request an Annual Quote

Mt. Sinai Asthma App Study Looks to Improve Engagement Amid Scant Genomic Data Sharing


NEW YORK (GenomeWeb) – Despite successfully enrolling thousands of asthma patients in a Mount Sinai study through the use of a smartphone application, it has been challenging to retain participants' engagement, according to an expert leading the project.

In particular, Eric Schadt, director of the Icahn Institute for Genomics and Multiscale Biology at Mount Sinai, told GenomeWeb recently that study participants haven't submitted their 23andMe genetic data as readily has he had hoped.

Health app-enabled study enrollment and longitudinal patient profiling, particularly when it involves genomics, is fairly new territory. And Schadt's group is still figuring out the best approaches to keep participants interested and submitting data, which was the goal of the Asthma Health app.

With the launch of Apple's ResearchKit in March 2015, Mount Sinai announced the availability of the Asthma Health app, wanting to test out the feasibility of using it to recruit participants who then could submit a variety of data about asthma symptoms, triggers, medications, and fitness. Using the app patients could electronically consent to partake in the study. The app could pull in data from third party devices allowing researchers to track things like patients' inhaler use and blood pressure, as well as the pollen count in areas they traveled based on GPS sensors.

Initially, there was a blockbuster response. Helped by the fact that Apple highlighted the Asthma Health app at one of its developer conferences, 50,000 people downloaded the iPhone application within two weeks of its launch, and ultimately, 9,000 people were enrolled in to the study.

Using traditional research approaches, where patients would have had to come into a facility to give consent and submit to tests and other evaluations, it would have taken four to five years to enroll 1,000 patients at a cost of several million dollars, Schadt said at a recent conference in New Jersey hosted by the biotech industry group BioNJ. "In this case we did 9,000 people in two weeks for effectively the cost of developing the app, which in this case, Apple helped pay for." The cost of developing the app was around $150,000, Schadt estimated.

However, he later shared with GenomeWeb that despite the robust initial response, it has been challenging to keep participants engaged over the past year. "I suppose there is good news and bad news," he reflected. "With the Asthma App, while we did get that initial 9,000 [enrollees], like everybody else we've definitely not solved the retention problem." The number of people that currently remain as active users in the app study is in the "low hundreds," he said.

In March, to much media fanfare, 23andMe announced a new ResearchKit module that would enable its customers to upload their genetic testing results within Asthma Health, if they were participating in that study. At the end of June, when Schadt spoke to GenomeWeb, he estimated that around 50 people had shared their 23andMe genomic data using the app.

The scant numbers may be due to a number of factors that are not immediately clear, such as if there were a limited number of asthma study enrollees who had been tested through 23andMe in the first place, or if there were reasons why some didn't want to contribute their genetic information through the app.

A 23andMe spokesperson couldn't say how many of its customers had joined the Mount Sinai study and contributed data through the app, but noted that the firm didn't advertise this to its customers. The firm's website, however, does inform visitors that customers can integrate their 23andMe data into large-scale medical studies, such as Mount Sinai Asthma Health and Stanford Medicine's MyHeart Counts. (Stanford researchers involved with the latter study did not respond to interview requests for this article.) 

"I wouldn't characterize this as a smashing success," Schadt said, reflecting on the number of people who have shared their 23andMe reports through the app. "But again, our aim in doing this was as much to learn what we can do, learn how to turn this crank, and see how people would respond. I kind of view it as a big success in that way."

Using the app, for example, Schadt and his team have been able to enroll patients across 50 US states and collect a variety of environmental data that might exacerbate the condition, such as pollution levels, pollen count, and traffic patterns, hundreds of times a day. "For example, … New Jersey and New York were the only two states in the country that had an appreciable self-reported cause of exacerbation due to anger," he quipped at the BioNJ meeting. "We didn't see that in sunny California."

But the early, health tech-enabled gains in enrollment were eventually weighed down by the rules of the traditional research paradigm. In the early stages of the app, his team was trying to make changes based on the feedback from users, but all those changes had to be approved by the institutional review board (IRB), and "we lost a lot of people," Schadt explained.

"All of those changes, we couldn't just make them on the fly," he said. "We had to make amendments to the IRB and that took weeks and weeks." His team is working on how to streamline and automate this part of the process, and deal more efficiently with IRBs.

They are also working on providing more meaningful feedback to users, since many didn't have the ability to automatically register the drugs they were taking or have Bluetooth-enabled inhalers that they could link to the app, and would have to manually input this information every time.

When it comes to genomics, one of the challenges is communicating to patients the benefits of getting tested and sharing data for complex conditions. "[Asthma] does have a genetic component but it's very complex," Schadt said. "It's not as though we can better risk stratify somebody based on that information."

One of the things Schadt's group had hoped to do with the help of the app was identify protective effects of asthma by identifying patients with penetrant genetic markers that should have caused a more severe phenotype but didn't. "With the small numbers coming in that's really not going to be possible," he said.

So, for now, Schadt's team is focused on figuring out how to better engage asthma patients by identifying "the stickier points in their health course." For example, the most active and committed users are those that have the severest forms of asthma.

Another approach would be to engage a population that has already seen the value of genomics. For example, Mount Sinai provides genetic testing to 120,000 people a year in the reproductive setting, and surely, there are asthmatics in this population.

"What if we leverage that as our growth hack engine, because every woman thinking of becoming pregnant or is pregnant should be tested this way?" Schadt reflected. "Then, [we can] leverage all that genomic information for other reasons than the reproductive health journey … and into the more common diseases."

Ultimately, Schadt is a strong believer that apps will connect patients, providers, and researchers in the healthcare enterprise of the future. The medical systems he envisions won't revolve around hospitals, but around patients. Individuals will be profiled in real time, closer to their homes, with the help of wearable and implantable devices, which will send data back into "a NASA-style command-and-control center processing all this data in real time and identifying when an individual has bounced off his good trajectory on to a bad one."

Schadt's future medical system will also be better at modeling health course trajectories because apps, wearables, and implantables will be able to bring in much more than genomics data, which, while fundamental to these models, is only one aspect of the puzzle. With only genomics, "you develop knowledge of the disease but not understanding," he said at the BioNJ meeting, noting the need to integrate "different dimensions of data," including DNA, RNA, proteins, metabolites, clinical, physiological, and consumer-acquired information.

"These models aren't simple, linearly ordered pathways," he said. "They're very sophisticated non-linear constructs that don't act in isolation. They act in concert, in a network of networks that defines the complexity of our system. We need to be operating at that level."

This might seem pie in the sky to some now, but Schadt assured, "this data is coming."

The Scan

Machine Learning Helps ID Molecular Mechanisms of Pancreatic Islet Beta Cell Subtypes in Type 2 Diabetes

The approach helps overcome limitations of previous studies that had investigated the molecular mechanisms of pancreatic islet beta cells, the authors write in their Nature Genetics paper.

Culture-Based Methods, Shotgun Sequencing Reveal Transmission of Bifidobacterium Strains From Mothers to Infants

In a Nature Communications study, culture-based approaches along with shotgun sequencing give a better picture of the microbial strains transmitted from mothers to infants.

Microbial Communities Can Help Trees Adapt to Changing Climates

Tree seedlings that were inoculated with microbes from dry, warm, or cold sites could better survive drought, heat, and cold stress, according to a study in Science.

A Combination of Genetics and Environment Causes Cleft Lip

In a study published in Nature Communications, researchers investigate what combination of genetic and environmental factors come into play to cause cleft lip/palate.