Data Accuracy from Wrist-Worn Wearables does Not Make the Cut
It is apparent that wearables, and specifically, wrist-worn wearables, have swept the consumer market: the International Data Corporation reported vendor shipments of 78.1 million wearable device units in 2015, and Gartner predicts the sale of smartwatches alone to hit 50.4 million units in 2016. The fact that the top five wearable vendors specialize in wrist-worn devices doesn’t come as much of a shock, considering their profusion at places like the gym and the workplace.
Wearable devices like FitBits and Apple Watches provide the general population with data that is easy to obtain and easy to understand. Access to individual health metrics motivates users to exercise and prioritize their fitness, but, on a larger scale, wearable technology that can record biometrics has been a breakthrough in the quest to understanding the workings of the human body. The gold standard for biometric data capture, up until now, has been cumbersome, inhibiting wearer movements and restraining data collection to a clinic or lab setting. In sleep studies, subjects have been tasked with attempting to sleep while hooked up to polysomnography machines, leading to both subject discomfort and data that reflects an atypical night’s attempts at rest instead of capturing a sample of the subject’s typical sleep patterns. Cardiac monitors are bulky, loud, and far from discreet, leading to wearer distress and the likelihood of reduced compliance.
Wearables present an opportunity to continue learning from physiology in a much less obtrusive, and less expensive, way. For researchers, this means more compliance, more accuracy, more data, and ultimately, more insights. Wearable devices are transitioning from helping people learn about themselves, to helping us learn about humans as a whole, shedding light on a variety of study areas including human performance, aging, and mental health. In March, MobiHealth reported on 21 ongoing clinical trials using FitBit trackers and then followed up with a second post in April covering 18 additional clinical trials incorporating the devices. There are currently 20 open and active cardiac clinical trials using wearable tools, according to a search on clincatrials.gov for the term ‘wearable’. In the pharmaceutical space, Biogen funded a study using FitBit as a tool to help patients track their activity, motivate an increase in mobility, and understand their Multiple Sclerosis. Earlier this year, Pfizer and IBM announced Project Blue Sky, a partnership built on the mission of revolutionizing Parkinson’s care with the aid of wearable sensors.
As wearables are increasingly penetrating the evolving healthcare landscape, shouldn’t we make sure all this data is as accurate as possible? The data recorded by wearables is informing the way people monitor their health, the way doctors treat patients, and the way drugs get to market.
Data from the wrist isn’t enough.
The benefits of wrist-worn devices are clear - people don’t mind wearing them. We’re used to wearing watches and bracelets, so as far as continuous monitoring goes, throwing on another armband doesn’t seem so bad.
The Journal of the American Medical Association published a study this October investigating the issue. The study, titled “Accuracy of Wrist-Worn Heart Monitors,” compared the Apple Watch, Mio Fuse, Fitbit Charge HR, and Basis Peak to the gold standard accuracy of the Polar H7 chest strap. Though the Apple Watch came closest, none of the wrist-worn devices matched the accuracy of the chest strap, and generally seemed to decline in accuracy as exercise and exertion increased.
As MIT Tech Review’s Rachel Metz points out, though the “wrist seems like a logical spot for a wearable device . . . it’s actually not a great place on the body to track biological signals.” She covered the subject back in 2015, in “The Struggle for Accurate Measurements on Your Wrist.” Nearly a year later, Metz expanded the investigation with an update on the startup Quanttus, and their failed efforts to “capture reliable, continuous signals at the wrist.”
In order to achieve the ideal fit for accurate heart rate readings, for example, armbands need to fit snugly, but still comfortably, around user’s wrists. Finding a fit that works for everyone is a tall order, as no two wrists are alike. Wrist captured data comes with the unwelcome bonus of motion artifacts and doubt.
Assuming a wrist-worn strap worked exactly as it should - is the wrist really the most ideal spot for capturing data? Valencell CEO, Steven LeBoeuf, doesn’t think so. As cofounder of the biometric sensor developing company, LeBoeuf proposes the ear as an alternate, and more accurate, source for recording heart rate. LeBoeuf shared Valencell’s reasoning with Metz, explaining that the ear provides a “stronger signal and less noise,” and since “we don’t move our ears as much as our arms, it can be easier to sort out intentional motions from unintentional ones.” Building a device small enough to fit in an ear, and versatile enough to fit any ear shape, is another challenging endeavor.
LeBoeuf’s heart rate innovations, and the standing supremacy of the chest-worn heart monitor, illustrate the reality that one spot can’t do it all. We can’t expect one body part to be the ideal sensing location for every physiological response. The wrist can’t wear every data capturing hat, but neither can the ear or the chest.
As the search for more accurate, more reliable data continues, and wearables play an increasingly more significant role in defining our health and lifestyle choices, researchers shouldn’t feel confined to swinging, irregular wrists. The specific study, and the physiological response in question, should define the data’s point of origin. It’s time to think beyond the wrist.