The United States National Institutes of Health (NIH) wants to tie development of mobile health apps to evidence-based research, and it hopes to do that with a new grant program. The imperative to align developers with research is urgent, given the strong interest in health IT, mobile health and health data. There are significant challenges for the space, from consumer concerns over privacy and mobile applications to the broader question of balancing health data innovation with patient rights.
To learn more about what’s happening with mobile health apps, health data, behavioral change and cancer research, I recently interviewed Dr. Abdul Sheikh. Our interview, lightly edited for content and clarity, follows.
What led you to your current work at NIH?
Dr. Abdul Sheikh: I’ve always had a strong grounding in public health and population health, but I also have a real passion for technology and informatics. What’s beautiful is, in my current position here as a program director at the National Cancer Institute (NCI), I have a chance to meld these worlds of public health, behavior and communication science with my passion for technology and informatics. Some of the work I did before coming to the NIH was related to the early telemedicine and web-based health promotion efforts that the government of Canada was involved in.
At NCI, I direct a portfolio of research on technology-mediated communication. I’ve also had the chance to get involved and provide leadership on two very cool efforts. One of them is leadership for our division’s Small Business Innovation Research Program (SBIR). I’ve led the first NIH developer challenge competitions as well.
What are some of the challenges that NIH has sponsored and the outcomes?
Dr. Abdul Sheikh: We’ve been very excited by the White House Open Government Directive and the authorization of America COMPETES in 2010. In the Department of Health and Human Services, within which NIH is based, we’ve also been very excited by the Health Data Initiative that Todd Park, the former HHS CTO and now U.S. CTO, launched.
It really feeds into a lot of the research that we’ve been funding in my division at the National Cancer Institute. We’ve been funding the development of applications and technologies for cancer prevention and control. To unpack that a little bit, cancer control involves this notion of the cancer continuum, which is prevention through detection of cancer, diagnosis and treatment, and survivorship. There are a lot of crosscutting issues that patients and providers face related to preventing cancer, in terms of behavior and diet, tobacco and smoking cessation, and informed decision-making.
When you get a cancer diagnosis screening test, how do you make a decision about how to move forward with your treatment? How do you talk to your provider about it? How can your providers convey technical information to you?
What we found is that when America COMPETES was launched, it gave us the opportunity to reach out to new groups of stakeholders that we hadn’t been reaching out to before to address these challenges of cancer control. Those groups include really incredible innovators, the data scientists, the developers, the entrepreneurs who are involved in this growing and emerging area of application development, of technology and informatics.
Back in 2010, we met with Todd Park and, after talking to him about the COMPETES Act, we got really excited and launched our first developer challenge in 2011. We were very pleased to partner up with ONC, the Office of the National Coordinator for Health IT. We put out a challenge with Health 2.0 called “Using Public Data for Cancer Prevention and Control: From Innovation to Impact.”
This was a challenge where we asked entrants to develop software apps that address some of the challenges that consumers, clinicians and researchers are facing across this cancer control continuum. We wanted them to use the publicly available data that we have here at the NCI, NIH and at our sister agencies in HHS. Examples would be improving health-related behaviors, like physical activity and smoking cessation, or helping to inform decision-making related to cancer screening or helping adherence to treatment plans. There are really a lot of areas that applications could’ve targeted. We were really pleased because we got around 20 applicants who came in with prototype apps. This was in the summer of 2011.
We were gratified to know that we were getting such a positive response from the developer community. They were working with our data to develop apps for cancer control. It was a really great process. One winner developed an app called “My Cancer Genome.” This is from Dr. Mia Levy at Vanderbilt. It’s an interesting application that provides technical information related to treating cancer with genetic information. It’s targeting oncologists. She really pushed the bar on what we expected to come out of these challenges by providing decision support and developing an app that has the potential to be linked and integrated with an existing EHR platform. After winning our challenge, Mia’s team went on to win the GE Cancer Data Challenge.
That’s just one example of the types of products that came out of the challenge and how they’re addressing both our broader goals at NCI for cancer control. They’re really allowing developers to flex their muscles and address difficult problems such as cancer.
What’s working as we look at behavioral change research around cancer? What isn’t? What can health data tell us?
Dr. Abdul Sheikh: One way to parse this is to look at consumer versus clinical settings. When we talk about the cancer continuum and prevention, I think about Lorien Abroms, who’s at George Washington University.
She recently did a review of iPhone apps for smoking cessation in the App Store [STUDY] which found that almost none of them were using what we already know works for helping people to quit smoking. These are behavioral and counseling and other techniques that can be used and potentially integrated into apps. This concerns us. We have a growing evidence base of science that shows you can focus on certain psycho-social factors, on environmental issues related to the context of the individual, to change their behavior for cancer prevention. It’s not being translated into the real explosion of innovative applications and technologies coming out there that use data and track user behaviors, like the quantified self movement. I would love to see the quantified self movement get more involved and have a deeper understanding of the evidence that we know for behavior change.
On the clinical side, we’re seeing that the restructuring of that service environment has created new decisional architectures for cancer care. These new environments have the potential for improving patient-centered communication and decision support and care coordination.
An example of the really interesting work that we’re funding can be found at the University of Wisconsin, where one of our top scientists, Michael Fiore, is the Director of the Center for Tobacco Research. He’s working with a team that’s developing EHR [electronic health record] decision support for smoking cessation. Using what we know that helps people quit smoking, he’s integrating and developing a system that can be put into EHRs for reminders so that clinicians will know, “Here’s a patient. Here’s what their stage is in smoking.” They can then give them targeted information to help them quit smoking and to help them adhere once they do quit. By providing this in an integrated way into emerging electronic health information environments, it can really go a long way to helping doctors and other care professionals help patients.
Does research show that mobile health apps are making any difference to patient outcomes through behavioral change?
Dr. Abdul Sheikh: Yes, definitely. We’re seeing a number of studies being funded that are looking at sensor technology and accelerometers to track where people are, how active they are for physical activity, and to track their behaviors in context. This notion of behavior in environmental context is really important. That would take into account multiple levels of data that include your physical activity, where you are in a geographical setting, and what sorts of resources are available for you, in terms of being physically active or if you have grocery stores near you. Are you in a food desert or not?
All of these multilevel factors and how technology and mobile devices can be used to address them is being explored and funded, not just at NCI but at NIH as well. There’s a strong interest in mHealth; every fall for the past few years, NIH has cosponsored an mHealth conference.
What are the most promising data-driven approaches that are getting improved outcomes for cancer prevention or treatment?
Dr. Abdul Sheikh: One of the challenges with behavioral medicine is that there are multiple ways to collect data on phenomena. That could be, for instance, self-efficacy for physical activity, which is your confidence and your ability to be physically active. The problem with measuring the same phenomenon different ways is that it’s really hard to harmonize the data. You’re going to have data that’s been collected in different ways, and it won’t really be valid if it’s integrated together.
Tim O’Reilly and others have talked about moving from a passive information environment online to an active user participation environment online as well as collective intelligence, open health and Medicine 2.0. One of the efforts we’ve been funding here is a platform called GEM, which stands for “Grid-Enabled Measures.” It’s an experiment for us. We started this four years ago.
This platform has two goals that relate. One is to promote the use of standardized measures tied to theory-based constructs for behavior change. The other is aimed at facilitating the sharing of harmonized data that results from the use of these standardized measures. If we can get our scientific community at NIH to start collecting data with controlled measures, then we can utilize that data, mash it up and connect it with other sources of data much more easily.
We have this challenge of legacy data that’s been collected with different methods, with different tools and instruments. We really want to change that moving forward. This GEM platform is an exciting experiment for us to engage with our community and get them to start thinking about using the same measures and sharing data, both of which are not very common in some scientific disciplines.
How have the ways that the NIH has used health data changed over time?
Dr. Abdul Sheikh: We’re very interested in visualization methods and technologies for making sense of data for various audiences. We have a proof of concept called PopSciGrid. We’ve worked on this concept for about four or five years now. We were really inspired by Hans Rosling’s first TED talk and how he presented complex data looking at outcomes related to health and socioeconomic status. He made it so simple to understand, and that really inspired us.
With PopSciGrid, we worked with two very smart sets of people. One is based over at Northwestern, in the SONIC lab. The other group is based at RPI, led by Deborah McGuinness at the Tetherless World Constellation. We gave them data related to smoking prevalence across the U.S. and data related to smoking policies.
Beyond the intellectual question of whether policy did have an impact on smoking, we wanted to really start looking at how can you convey such complex and different datasets. We had data from the CDC, from the NCI and from some other sources. How could we mash it together and make it available in a way that gives new insights not just to technical audiences, like scientists, but also to consumers and to policy makers?
Over the course of 10 years, as the smoking cessation policies got stronger, we looked at how that impacted smoking cessation and where it impacted smoking cessation across the U.S. Both Deborah and Noshir [Contractor]’s team helped us develop really great maps where we can see how smoking rates have changed with the introduction of these policies.
What should readers know about this new SBIR grant program and how to get more involved in the health information technology industry?
Dr. Abdul Sheikh: We announced the program in June at the Health Data Initiative Forum in D.C. We’re very excited about this. Very briefly, the function of the SBIR funding mechanism is to commercialize products that are evidence-based. It’s all about giving your community money to develop applications and technologies for cancer control.
These would include preventing or reducing the risk of cancer, facilitating patient-provider communication, improving disease outcomes in consumer and clinical settings. It’s quite broad.
We’re looking to fund the development of apps and technologies that are evidence-based. By that, we mean do they really hit the psycho-social predictors and the pain points that our research is showing will lead to improvements in cancer-related outcomes.
We’re excited about this SBIR funding opportunity because we’ve already had a wonderful response from the innovation community with our two developer challenges. With SBIR, it gives them potentially $1.15 million and three to four years to further develop an application or technology for cancer control. It allows them to then keep the intellectual property and to commercialize it as they see fit.
There is a bar to entry because we have peer review for the applications that come in. We’re also asking these small businesses to come in with a large business partner. That would include commercial IT firms, EMR or EHR vendors, health care systems and hospitals, and not-for-profits. A small business applicant coming in with a partnership with a large business is important because it’ll show the reviewers that you’re working with this large business to integrate your app or technology into their platform. They can help you develop it. They can provide a user base to evaluate it.
We’re excited because this also addresses the challenge that the Abroms paper found, which is that a lot of these health-related applications for behavior change are not translating to the evidence-based science that we know can actually help improve behaviors for cancer and other diseases.
Strata Rx — Strata Rx, being held Oct. 16-17 in San Francisco, is the first conference to bring data science to the urgent issues confronting health care.
Photo Credit: National Cancer Institute