8 lessons Apple is learning from 8 mobile health apps
After first telling folks nearly six years ago that "there's an app for that," Apple is now working with big name university medical schools and development partners on apps for eight different maladies. Many thousands of iPhone users have already downloaded these apps from Apple's App Store. Meanwhile, Apple and its prestigious partners are learning valuable lessons about the types of features iPhone owners will happily use in the interests of getting healthier.
[ Related: ResearchKit at 6 months: 100,000 people now using medical apps ]
Apple and its university colleagues released the first set of these medical apps -- for asthma, Parkinson's disease, diabetes, breast cancer, and heart disease -- in March of 2015, following on in mid-October with new apps for autism, epilepsy, and a potentially deadly form of skin cancer known as melanoma.
All of these apps are built on Apple's ResearchKit, an "open source development framework" that lets software developers create disorder-specific apps that work with the iPhone's HealthKit and hardware sensors aboard the phone.
In the Parkinson's disease app, known as mPower, patients can use the iPhone's sensors to take a series of interactive tests for measuring and tracking their memory, speech, motor, and gait/balance skills.
With the Autism & Beyond (A&B) app, parents use the iPhone's camera to shoot videos -- snapped with the kids held in their laps -- that can help with early screening for autism.
"Although ResearchKit is not without its limitations, Apple is to be commended. There's never been anything like ResearchKit before, and it has the potential to change existing ecospheres. Apple's ResearchKit partners are trailblazers, and you can be sure that they're getting back to Apple with some of the lessons they're learning,” said Pam Baker, an industry analyst and consultant, in an interview with ITworld.
Here are eight such lessons Apple and its research partners are learning from the eight apps.
“Asthma Health needed to be as ‘sticky’ as other popular apps in order to generate the required amount of study data,” Corey Bridges, CEO of software development firm ) LifeMap Solutions, said in a blog post on Apple’s ResearchKit site back in May.
“It’s one thing to get users to download your app, but for your app to succeed, you need users to return to the app the day after they install it, then after one week, one month, etc.,” Bridges told ITworld.
“Social networks and games generally have higher retention rates than business or health apps. Based, we believe, on our use of compelling features, good design, and timely reminders, we’ve been able to keep users coming back to Asthma Health,” Bridges said. When users of the app are sent reminders each Monday, usage spikes.
Successful mHealth apps will need to continually engage participants, concurred Brian Bot, principal scientist at Sage Bionetworks, co-developer of the mPower app with the University of Rochester. In mPower, tasks where patients are asked to engage in an activity tend to be the best received and most often completed.
“Not surprisingly, saying ‘aaaaaaah’ into a phone for ten seconds or trying to tap as fast as you can carries much less friction than completing a 20-question survey,” Bot told ITworld in an email.
ResearchKit’s main limitation is that the first crop of apps runs only on Apple hardware, according to Baker. Moreover, some researchers maintain that the data from the ResearchKit apps is skewed, since iPhone users tend to be younger and much more affluent than the overall US population.
Five of the ResearchKit applications are open source, an advantage designed to allow ports to Andoid as well as integration with existing devices such as medical monitoring systems.
Apple is reportedly trying to encourage Android ports by developers.
“We are actively interested in [Android ports] ourselves,” said LifeMap’s Bridges. “We haven’t revealed our schedule, though.”
Yet, although these ports are possible, they’re not a proverbial piece of cake, some partners say.
“The difficulty will be reconciling the differences between sensors such as accelerometers, gyroscopes, microphones and cameras (on various Android phones),” according to Bot. But “there are efforts moving forward among the Android community,” he added.
“More than 50,000 people downloaded our app," recalled Dr. Yvonne Chan, director of digital health and personalized medicine at Icahn School of Medicine, in a summary of preliminary results for the Asthma Health app. Of those 50,000 downloads, 8,800 people “met our strict eligibility criteria to join the study (asthma diagnosed by a physician, on prescription meds for asthma, USA resident, 18 or older),” according to Chan.
“We broke the geographic barrier that typically limits traditional research to the local area of the university/medical center. For our study, 87% of participants live outside of NY and NJ!”
This kind of reach could also come into play for scenarios involving remote diagnosis by medical specialists based in far off locations.
Various other analyst firms, many of them based overseas, have pointed to rapid future growth in regions like Latin America and Asia-Pacific. They cite factors such as rising penetration of mobile devices and increasing awareness that remote monitoring of chronic illnesses can help to cut re-hospitalization and other healthcare costs.
The international market opportunity can’t have escaped notice from Apple, which is teaming with Duke University on an autism app created in three languages: English, Spanish and Chinese.
The English version hit Apple’s App Store in October. The Spanish version is slated for availability within the next month, with the Chinese version set to follow shortly afterward, according to Guillermo Sapiro, professor of electrical and computer engineering at Duke.
The Spanish and Chinese versions now need “careful translation to the local language [and] institutional approval from the IRB (institutional review board). From the technical perspective, all is ready,” Sapiro told ITworld.
Autism researchers at Duke wanted to build an app that videotapes a child's face as the child interacts with the phone. They were finding, though, that the processors in older iPhones couldn't keep up with 30 frames-per-second (fps) video.
After a week of meetings with Apple at the software giant's Cupertino, CA headquarters, the Duke researchers returned to their labs. Testing by the university showed that the final app works fine on the iPhone 4 and higher for meeting Duke’s research goals, according to Sapiro.
“Note that from the scientific perspective, there is no clear indication that 30 fps video is needed for [autism] screening. Clearly that is not the rate at the clinic, where humans perform the observations with the naked eye," he told ITworld.
Johns Hopkins’ EpiWatch app is aimed at collecting data from epilepsy patients before, during, and after their seizures in the interest of improving seizure detection, medication adherence, and quality of life for people with epilepsy.
The app includes modules that run on the Apple Watch as well as the iPhone. According to the researchers, the app works best among patients who experience an “aura” signaling that a seizure is approaching. At that point, the patient or a caregiver can tap a button on the watch to activate the app.
Professors at the university’s medical school worked with THREAD Research to build an app that uses sensors on the watch to record heart rate, movements, and falls. The watch also requests patient participation in a memory game to gauge the patient’s responsiveness during a seizure.
In a “sticky” feature calling for ongoing engagement with the app, users keep daily journals about their seizures and medications.
On the backend, the app uses a solution from Acuma Health, a division of Smart Monitor, to protect patient data and supply researchers with custom dashboards.
In a study about personal health data (PHD), the Robert Wood Johnson Foundation (RWJF) found that 75% of survey participants said they either "probably would" or "definitely would" share their information with medical researchers, Bot said during a presentation at a recent mHealth conference in New York City.
Meanwhile, after receiving enthusiastic response from asthma sufferers, LifeMap plans to test out a theory that Asthma Health users are motivated by the goal of participating in a project that will help the entire asthma community.
“We’ve gathered more anecdotal data on that, to help hone our hypotheses, and we have some additional ResearchKit apps planned, which we can use to test how other groups’ motivations compare to Asthma Health’s users,” Bridges told ITworld.
Also according to the RWJF’s research, though, more than 90% of users are interested in keeping their health data anonymous, said Bot.
Sage Bionetworks is complying with these kinds of wishes. Data from mPower and the other four original ResearchKit apps are using the same backend systems from Sage.
“The apps communicate directly with our Bridge Server APIs (application programming interfaces), which separate account information from the coded research data. This study-specific coded research data is consolidated for each task and exported to Synapse, providing researchers daily updates and a queryable interface into the over 2.5 million records collected across the five apps,” he told ITworld.
Through this process, the names of participants and other personal information are removed from the data that researchers can see.
Users who sign consent forms agreeing to participate in any of the first five ResearchKit projects can choose to share their anonymous data either "with Sage Bionetworks and its partners" or "with Sage Bionetworks and its partners and qualified researchers worldwide." Most mPower users decide to share their data more broadly, he said.
“We (Sage) certainly plan on making data from the mPower app and subsequent analytical results and insights available both through traditional methods (publication) as well as data releases for the subset of participants who consented to share their data broadly with the scientific community," according to Bot.
"For all five of the apps, only partners explicitly listed on IRB submissions as research 'partners' can access the full set of data. For the subset of participants who have agreed to share broadly, any qualified researcher who has agreed to a set of use conditions will be able to access this subset data."
Bot also maintained, however, that members of Apple’s executive team have assured Sage that Apple will not access any of this data.
Meanwhile, in September, IBM Watson Health and Sage Bionetworks agreed to work on an “open biomedical research platform” made up of the Bridge Server and Synapse technologies powered by Watson Health Cloud and Analytics. “The implementation is currently being worked out,” Bot told ITworld.
“I think the market is probably significantly underestimated," said Apple CEO Tim Cook, in a televised interview soon after Apple’s announcement of its first set of ResearchKit partnerships.
In fact, the mHealth technology market is changing so quickly that it’s hard for market projections to keep pace. However, in a detailed five-year roadmap published last year, industry analyst firm research2guidance projected that the mobile health apps market grew from about $2.5 billion in 2012 to $6.4 billion in 2014 and will reach $26.6 billion in 2017. The analyst firm pointed to big changes between 2012 and 2014, as perceptions of mHealth became increasingly businesslike and the market entered the commercialization phase.
Developers are using iOS and Android to target both the fitness and chronic disease markets in a rapidly growing way, with the number of apps doubling in two-and-a-half years to reach 100,000 in the first quarter of 2014.
Still, mHealth developers represent a very mixed bag, ranging from small mom-and-pop shops with revenues of zero to $10,000 to behemoths raking in $1 million or more annually through downloads, in-app purchases, services, and/or sales of sensor-driven hardware devices such as wristbands, scales, or blood pressure units.
By 2017, the majority of income for mHealth app publishers (69%) will come from services. “These services typically involve backend structures of servers and/or teams of medical staff which monitor and consult with doctors, patients and general healthcare-interested individuals,” according to research2guidance’s report, which is based on surveys of developers of all sizes.
The two app categories with the highest expected market potential in the near future are remote monitoring and consultation.
The analyst firm foresees increased market acceleration as early as 2016, when the mHealth market will have entered the “integrated market phase,” featuring integrated healthcare solutions together with first-time payments for mHealth services by insurance companies.
ResearchKit apps, which are largely about experimenting with various features for remote disease monitoring and diagnosis, aren't even in the category of commercial apps – not yet, anyway.
“This initiative from Apple will [not] have a major impact on the market in the next few years,” says Ralf-Gordon Jahns, managing director of research2guidance. Rather, he asserts, ResearchKit “is more of a [toolset] for helping to provide evidence for medical outcomes than for directly generating revenues and downloads.”
Still, though, the lessons learned from ResearchKit apps will likely inform and influence the design of other categories of mHealth apps, many of which could generate huge revenues, indeed.