Imagine a fitness tracker like a Fitbit, Up, or Misfit Shine on every person on the planet. Imagine all that data flowing up into an anonymized cloud-based platform. Imagine connecting that with health data from medical sensors. With location and altitude data from smartphones. With health information such as diseases we catch, conditions we develop, and accidents we encounter.
Imagine what you could learn about health and how to prevent disease.
“This would be the biggest imaginable pool of people for researchers to study,” Dr. Richard Hu told me last week. “It’s an opportunity to create native information that is out in the wild that doesn’t get biased or influenced by study parameters.”
Hu’s company, Vivametrica, is building a cloud-based platform for aggregating data from any kind of bio-sensor, from popular fitness trackers to medical-grade health sensors. Vivametrica plans to store that data for you so you can see all the data from all your devices in one place, compare it to others who are like you, and eventually feed that into medical applications that doctors and HMOs could use to help prevent illness, diagnose conditions, track health, and discover unprecedented insights into what truly constitutes healthy living.
And not just just to derive averages. Hu expects to be able to provide specific, actionable health advice for you, based on others who are similar in age, sex, health status, fitness level, geographical location, and more.
“There’s a cloak of digital data around all of us. We want to take advantage of that and give people tools to better manage their lives and activities,” he said.
That’s the first step: a personal fitness platform that derives value by virtue of, essentially, a new kind of streaming-data wisdom of the crowd.
To feed it data, Vivametrica is building an analytics platform that receives data from dozens of incompatible fitness trackers and health monitors, as well as smartphone data like location or even altitude. Apple’s recent health initiative, HealthKit, “makes our job exceptionally easier,” Hu said,” adding that the company is “able to bring anything that goes through an Apple device into our system.” Vivametrica is working on similar aggregation technology for the Android universe, using the Google Fit and Google Wear frameworks. The company will support any phone using KitKat or above, which will be 75 percent or more of Android phones by the end of next year.
As the platform grows, however, the possibilities get even more interesting.
As much as the fitness-tracking industry is diverse, it’s a model of a consolidated industry compared to the medical devices industry, where there’s much more fragmentation in terms of both operating system and proprietary data architecture. This is already a significant challenge in the use of these devices in medical settings, Hu says. So Vivametrica will start to onboard those on the platform as well, starting with the most widely used blood pressure monitors.
The vision is that with hundreds of thousand or millions of consumers on board, Vivametrica would provide better, more accurate, and higher-fidelity data of what people actually do and what health conditions result than almost any prior health-focused scientific study — most of which are lucky to have participants numbering in the hundreds.
With an ever-growing data pool like that, the algorithms should just get better and better. And the integrations into medical data systems would become ever more attractive.
“There is a space out there for good quality information that health care providers will welcome and [that will] add value to patients,” Hu said. “The ultimate plan is an SDK for the platform so independent developers can design apps to use the data and perform analysis, keeping in mind privacy and security. If we provide the primary tools and platform … those people who find this to be useful will start to design and use this in appropriate applications.”
The challenge bedeviling non-medical-grade health devices, of course, has always been whether doctors can trust the inputs. I know from personal experience that FitBit, BodyMedia, Up, and the Misfit Shine can differ significantly in terms of how many steps they think I’ve taken, for instance, and how hard I’ve worked out.
But Vivametrica has a plan for this: data normalization painstakingly developed via testing all the most popular devices, and cross-referencing them to the gold standard in the medical industry for activity tracking, the ActiveGraph. The ActiveGraph has been used in scientific studies, has very defined and calibrated accuracy, and has very structured software, Hu said.
“We found the Nike FuelBand correlation coefficient is .89 [to the ActiveGraph], which implies there is a 10 percent error,” he told me. “However, the Misfit, the Jawbone Up, and the Fitbit are at .94, .96, and .97, so they are in fact relatively accurate and reproducible in terms of step count.”
So the company has simply developed a constant variable for each major device, mapping it to what the ActiveGraph would have read … essentially, cleansing the data in a way that’s retroactively changeable as the comparisons get more and more accurate.
Ultimately, this kind of platform could be useful for a wide range of applications in personal and regional health. It could also be used to calculate life insurance premiums, track compliance with “doctor’s orders,” or provide national data on health and mortality.
That is, of course, if Vivametrica achieves its goals.
The company has been bootstrapped until recently, when it started raising a seed round. Clinical trials begin in a few months, company president Scott Valentine said, and a private beta will begin shortly.
The question is whether giants like Apple or Google, both of which have health initiatives, will help or hinder. In Valentine’s opinion, pretty much everything they do around health, like the recently-unveiled Apple Watch, can only help.
“I look at it and I see millions of devices that can help drive our platform,” he said.