Opinion: Wearable device data need to represent all of us

The growing uptake of wearable devices has sparked enthusiasm among health researchers, and for good reason. The data generated by fitness and activity trackers offers an exciting new stream of metrics — physical activity, sleep, heart rate, and more — that may help researchers better understand the unique health needs of specific populations.

But the promise of wearable data comes with a familiar pitfall. The use of wearable devices is currently concentrated among young, white, college-educated, and higher-income users which could worsen the problem of underrepresentation in research without intentional work toward equitable access to and use of digital health technologies.

Lack of diversity among those participating in medical research has created knowledge gaps and fueled disparities in health outcomes. For example, hospitalization and death among children with asthma vary starkly by race and ethnicity. By including children from racial and ethnic minorities in their large whole-genome sequencing study on drug responses in children, University of California, San Francisco researchers identified an important clue to these disparities: a genetic variation that could make albuterol, a common asthma treatment, less effective in Black and Puerto Rican children than in children of European or Mexican descent.


Putting wearable data to the test

The National Institutes of Health’s All of Us Research Program, which we are involved in, has been working to help address long-standing gaps in research by building a health research resource that reflects the diversity of the U.S. About 80% of the program’s participants come from communities historically underrepresented in medical research, including about 50% of participants who identify as a racial or ethnic minority. These communities are often left out of biomedical research. For example, more than 90% of participants in genome-wide association studies to date have been of European descent.

Recognizing the value of real-time data derived from wearable devices, especially when combined with the growing repository of genomic, electronic health record, and participant-reported data, All of Us enabled program participants to contribute data from their own wearable devices — beginning with Fitbit devices — starting in 2018. The program’s secure data analysis platform includes records that have been stripped of personal identifiable information, for wearable data contributed by more than 11,600 participants. The wearable data is linked with each participant’s survey responses, electronic health record information, and physical measures, representing one of the largest and most diverse publicly available sets of digital health technology data to date.


Given the user demographics for wearable technologies, All of Us recognized early on in collecting this information that it would not likely match the program’s goals for diversity. When our team compared the demographics of all program participants to Fitbit data contributors, we saw the representation of historically underrepresented communities shrink.

To understand the barriers, All of Us turned to an essential link in the program’s efforts to reach diverse participants: federally qualified health center partners. These are community-based health care organizations that provide primary care in underserved areas. They play a crucial role in the success and progression of All of Us, particularly in engaging communities historically underrepresented in medical research.

Six of these centers surveyed 1,000 people from communities historically underrepresented in medical research. As we and colleagues describe in the journal npj Digital Medicine, respondents who were interested in using an activity monitor made it clear: their lack of use wasn’t because they were unwilling to use devices or contribute the resulting data to research. They cited cost barriers (49%), the need for support in using the devices (19%), and a lack of a clear understanding of the potential value of these devices to their health (16%).

Shrinking the access gap

The findings from this survey point to opportunities to overcome this challenge and, as past efforts have shown, that underrepresented communities are not out of reach. All of Us is now piloting an effort to distribute wearable devices to examine ways to increase use among participants, and to educate, engage, and encourage participants about using these devices regularly. As we write this, about 3,000 All of Us Research Program participants are sharing data from program-provided devices, nearly all of whom are from communities historically underrepresented in biomedical research.

As digital health technologies become increasingly important sources of data in biomedical research, the research community needs to double-down on working toward inclusion and equity across the research ecosystem. That requires a sustained commitment to work with community partners and an investment in staff and funding for education, awareness and engagement. Researchers can analyze the data emerging from such efforts to show what is successful and what isn’t, and be ready to pivot, adapt, and take action.

In doing so, the research community can overcome existing gaps in representation and ensure new disparities are not created. It’s clear that the willingness of participants to adopt wearable technologies is there. Purposeful efforts to build access, opportunity, and infrastructure will allow all communities to benefit from the scientific knowledge gained through wearable technology data.

Yashoda Sharma served as a principal investigator for the National Institutes of Health’s All of Us Research Program at Community Health Center, Inc., a federally qualified health center in Middletown, CT, and is the program director for the Digital Medicine Society. Chris Lunt is the chief technology officer for All of Us.

Source: STAT