By Laura Lovett, MobiHealth News | May 10, 2019

The digital health industry has much work to do when it comes to developing technologies for underserved populations, including people of color, the LGBTQ community and women.

It’s no secret digital health is iterating and evolving at breakneck speed. But many are starting to ask the question, “Are those technologies made with all in mind?” Industry players have warned that many patients groups are being left behind— namely, people of color, underserved communities, members of the LGBTQ population and women.

The problems and opportunities 

“I would say communities of color, and frankly underserved communities of any color, are being left out of this conversation and the technology shifts,” Silas Buchanan, CEO of the Institute for eHealth Equity, told MobiHealthNews.

But there is hope. Buchanan said the emerging focus on environmental and social factors in healthcare has put a new spotlight on meeting the needs of every community.

“With this increased focus on social determinants of health, there will be much more attention given to those populations and maybe the ideators and the innovators will be more thoughtful to those populations,” Buchanan said.

Developing technologies for patients who might have previously been ignored also makes business sense. For example, the Medicaid population is quite large.

“Given the size of Medicaid, it covers — depending on what point you are counting — 17 million Americans, obviously all living in low income. A lot of them children,” Vanessa Mason, research director at Institute for the Future, told MobiHealthNews. “Look at the breadth of health needs, combined with a lot of the social, behavioral and environmental challenges that a lot of the [patients] covered by Medicaid [experience], especially with the expansion. I think it has been under invested in, and under represented in … the digital health space.”

So why aren’t more entrepreneurs looking at this population? Mason said she’s heard a slew of excuses around why folks aren’t investing and innovating for the Medicaid population.

“I’ve heard everything from ‘patients on Medicaid don’t use technology,”” Mason said. “I’ve also seen, on the market opportunity side, ‘There is no one who going to pay for that.'”

However, there are successful examples of digital health companies catering to Medicaid patients, according to Mason. She named Wildflower Health, a digital health platform that helps patients navigate benefits and connect to resources, as an example of a company working with soon-to-be and new moms enrolled in Medicaid.

Beyond Pregnancy

While Wildflower is an example of a company designed to address women’s health, there is still a gap in technology developed for women’s health. The femtech industry may be on the rise, but more than 60% of the tools on the market are centered around fertility, periods and pregnancy. Mason said she is encouraged that more companies have entered the women’s health space, but would like to see it address women’s needs across a continuum of life.

“We have this very defined paradigm of pregnancy in this country that doesn’t really include populations like lesbians and trans women,” Mason said. “So the question I want to leave you with is, what does designing for wellbeing and health look like for all women; specifically, what does that wellbeing look like when it doesn’t include pregnancy?”

There are more tools coming into the market that deal with menopause, she said. In December, for instance, VRHealth unveiled its new virtual reality product aimed at curbing hot flashes, called Luna.

Mason also cited work on technology to help identify biomarkers for chronic fatigue syndrome, a condition that disproportionately impacts women, as a positive step in the right direction.

“I think that is really emblematic of a lot of the conditions that women deal with, these invisible disabilities, invisible disorders that take forever to diagnose but cause a lot of suffering and poor quality of life,” Mason said. “I think partially why diagnosis takes so long is that it is trumped up to women [that] it is all in your head, or it is not being taken very seriously. There is generally this halo of women not being taking seriously within the realm of their healthcare, so I’m really excited to see what are the opportunities to take women’s health seriously.”

Can tech curb medical bias?

It isn’t just innovating and investing where bias comes into play. Across the medical profession, clinician bias has been a concern.

In fact, a systematic review of doctor’s racial bias on clinical decision making, published by the Academic Emergency Medicine Journal in 2017, found that an implicit racial bias for white patients was common among providers, regardless of speciality (though it’s important to note that this same study found a much weaker relationship between doctors’ implicit racial bias and their decision making).

“In medicine you are kind of taught to stereotype. This is the problem,” Dr. Damon Tweedy, whose autobiography,  Black Man in a White Coat, is about being African American male doctor, said at The Atlantic conference in Boston. “You are given this snippet and then have to figure out what the issues are. In medicine you start with this issue of what race, age and gender — so, a 30-year-old black woman. Then there are all these assumptions that go into that. Medicine is taught in this really biased way, which is really, really problematic and is something I think is often overlooked.”

Technology could have the opportunity to close some of this human bias, but it could also widen it. Mason said it comes down to how that technology is developed. She gave the example of clinical decision support tools.

“Clinical decision support tools are going to be built on top of already known clinical protocols and possibly known clinical behaviors,” Mason said. “Think about it from that perspective, to the extent that evidence-based medicine makes sense and we have accounted for disparities in care and include randomized clinical trials, and hopefully these clinical trials are included so that they are representative of the population and we have not built in biases through the addition of digital technologies and clinical support tools.”

However, technology is designed, developed and fed data by humans, which could pose its own set of issues.

“So if we are using that past behavior to then be able to inform AI machine learning models, we are encoding biases in these tools and they are going to make biases worse,” she said. “So it is really about getting back to the data that we have, and understanding how that data was analyzed, who was included or was adequately included, and then really looking at … the results coming from that.”

Bias is a hot topic within in the AI and machine learning field. Just this week at Google I/O, Google CEO Sundar Pichai hinted that the company is seeking to address this.

“Bias has been a concern in science long before machine learning came along,” Pichai said during the event. “The stakes are clearly higher with AI. It’s not enough to know if a model works, we need to know how it works. We want to ensure our AI models don’t reinforce bias that exists real world. It’s a hard problem, which is why we are doing fundamental computer science research to improve the transparency of machine learning models and bias.”

In fact, conversations around AI bias have been particularly prevalent in dermatology apps. While the benefits of clinically-validated automated screenings are easy to imagine, an article published in JAMA Dermatology last Augustsuggests that the training methods for these algorithms may not lead to accurate diagnoses among patients with skin of color

Pichai zeroed in on this particular problem during his speech and the company’s ambition to curb this problem in the future.

“Now imagine an AI system that could help detect skin cancer. To be effective it would need to recognize a wide variety of skin tones representative of the entire population,” he said. “There is a lot more to do, but we are committed to building AI in a way that is fair and works for everyone, including identifying and addressing bias in our own ML models and sharing tools and open data sets to help you as well.”

Where to start

Breaking down barriers between innovators and the patient communities is key, according to Buchanan. His organization teams up with secular and nonsecular groups to bridge this gap.

“I think people are being left out because there hasn’t been a very clear guided path to reach large numbers in any targeted way [for] community members of color,” Buchanan said.

A key way for clinicians and developers to get insight into different patient communities: hire diverse teams to help you innovate, he said. “There is a lot to be said for the lived experience,” he said. “You can know how people might think or feel, you can even conduct a focus group. But I don’t think anything [beats] having people on your team that represent a culture or racial or ethnic group that you are targeting with a health information technology intervention. There is no substitute for that.”