Americans collectively spend more on pregnancy and childbirth than any other nation yet have the worst maternal and infant mortality rates in the developed world. The U.S. is the only developed country in which maternal mortality is rising. And as the Centers for Disease Control and Prevention reported recently, the U.S. has seen increases in preterm births for four consecutive years as well as alarming rates of pregnancy-related deaths among black and Native American women.
It’s a complex problem made worse by a lack of maternal health data, scant research funding, and fear that commercial and public investments in anything pregnancy-related are too risky. According to a position paper by the American College of Physicians, women are underrepresented in research studies and, as a result, “progress is lacking” when it comes to understanding the causes of and solutions for many women’s health issues, including pregnancy-related deaths and health problems.
Data sourced from consumer health devices can accelerate clinical research and help fill this gap in medical knowledge faster than traditional methods. But this consumer-generated data must be held to standards used in medical research.
A paper published in January 2019 in the European Journal of Clinical Investigation found that so-called health care “unicorns” — startups worth more than $1 billion — were underrepresented in peer-reviewed research. More than half of current unicorns and nearly 40 of exited unicorns had no “highly cited papers,” and just two of the 18 current health care unicorns accounted for almost half of all unicorn papers. The authors concluded that, “Many unicorns may be overvalued and subject to unrealistic scientific expectations.” Indeed, some startups lacking the expertise to collect medical-grade data have nonetheless published misleading studies and developed products that left their users worse off than before.
The medical community needs more research and better data to address complex issues like maternal health. If clinicians, researchers, and innovators embraced standards like the ones I propose here, I believe we can ignite innovation in struggling medical fields, maternal health included.
The elephants in the room
For five years, my company, Bloomlife, has worked to simplify complex pregnancy issues, such as how to detect the onset of labor outside of a clinical setting. Toward that end, we developed a wearable sensor that more than 10,000 women have used to track contractions during their third trimesters of pregnancy.
We recently demonstrated in both laboratory and real-world settings that this device can identify when labor has started based on uterine muscle activity and parameters of the mother’s heart, with sensitivity and specificity equivalent to doctors using clinical exams. This device demonstrates the potential of consumer-generated data.
Naturally, this discovery has turned some heads in the obstetrics community. Current medical guidelines to diagnose labor require contraction monitoring and a pelvic exam to evaluate cervical dilation and effacement, yet we assert that our device can detect labor without either. Who are we to make this claim and why should our data be trusted?
It’s a fair question because consumer health companies do not have a perfect record when it comes to scientific integrity. Makers of fertility apps, for example, have made false claims about preventing unintended pregnancies. Theranos misled top investors — and even Walgreens — into trusting its highly inaccurate blood-testing technology. In the “quantified self” space, there are legions of startups promising to “empower” consumers with data. As both Reuters and the Wall Street Journal have reported, many such startups collect data and then pass it to advertisers and other third parties without the users’ knowledge.
There’s no shortage of ideas for how to revolutionize health care, but most fall short of revolutionizing anything, often at a cost to both patients and clinicians. Thus, the medical community should scrutinize startups that promise to revolutionize medicine — including mine.
A standard for trust
We are at the beginning of a massive shift in health care enabled, in part, by low-cost sensors, wireless connectivity, cloud computing, and artificial intelligence. When individuals’ health is on the line, a startup’s fake-it-till-you-make-it approach can have catastrophic consequences. Done responsibly, however, connected health can transform outcomes not just by empowering patients to better care for themselves but by aggregating longitudinal health data to improve doctor-patient communication and seed discoveries of digital biomarkers for breakthrough screening and diagnostic tests.
None of that is possible unless data can be collected in ways that meet medical standards and serve a purpose.
The strength of consumer health devices is also the source of their weakness. Although devices like Bloomlife’s labor sensor can gather sorely needed data points in private settings like the home, clinicians can’t confirm that people are using these devices correctly. The responsibility for data integrity — ensuring that data is trustworthy, secure, and used as intended — belongs to the connected health company.
I propose a standard for consumer health devices that makes data integrity a core focus of product development, user experience, and data analysis. This standard is meant for “the wild” — places where providers can’t supervise users. It goes as follows:
Field advancement. The data you intend to collect serves a purpose in academic research, clinical practice, and patient education.
Benchmarking. Your device has been tested against the gold standards used in laboratory or hospital settings.
Usability. The design of the device eliminates or minimizes user errors that jeopardize data integrity.
Signal quality. Your system separates good data from bad data in real-world environments where poor signal quality is at times unavoidable.
Outcomes. You collect reliable data on the outcomes people experience as a result of using your technology.
If data integrity is the starting point for good science that can advance neglected fields of research — and perhaps improve health outcomes — then consumer-generated data needs to meet a standard that both clinicians and researchers endorse. The standard I outline is a good starting point.
Better data, more breakthroughs
Although some innovators make unwarranted claims about their technology and data, putting the public at risk, many startups can and do meet standards cherished by the academic and medical community.
Consumer-generated data could lead to incremental discoveries that produce breakthroughs in reducing maternal and infant mortality. Other fields, including sleep medicine, mental health, and genomic medicine, are also positioned to benefit. For that to happen, though, we need startups to create technology with data integrity as their priority.
Clinicians, researchers, and innovators need to collaborate on data integrity now, before bad consumer-generated data undermines science and the medical field in general. Let’s work on developing trustworthy datasets that can address the world’s most complex health challenges.
Eric Dy, Ph.D., is the co-founder and CEO of Bloomlife.
