The Rise of the Wearables

 , January 30, 2015

A large number of “wearable” devices have been developed that can track various data about the wearer like activity levels, quality of sleep, heart rate etc. As these become cheaper, more ubiquitous and more capable, they will force medicine to reconsider how we define common medical conditions.  The amount of data generated by these devices can be overwhelming and be beyond the capacity of busy clinicians to interpret.  As data for multiple parameters becomes available, new patterns may emerge that may fundamentally alter how we think about medical conditions and how we can predict their course.

The recent report by the USPSTF is a case in point showing how ambulatory BP monitoring is changing how we define hypertension.  The report recommends that elevated isolated in-office blood pressure readings should be confirmed by ambulatory blood pressure readings before diagnosing HTN. This recommendation is driven by data that ambulatory blood pressure monitoring better predicts cardiovascular events than office blood pressure readings.  This makes common sense - when we can have data from "normal" settings over long periods of time, it gets closer to the truth than an isolated reading in a doctor’s office.  

As wearable devices become multi-functional and smarter, they will be able to track multiple parameters like activity, sleep, heart rate, oxygen saturation, skin temperature, calories burned etc.  Such devices can generate as much as 50 GB of data per person per year.  Such data streams can inundate clinicians and will require the help of computers for interpretation.  Machine learning has advanced rapidly in recent years and allows computers to recognize patterns in big data and interpret this data for humans.

With the increase in amount of available real-time data from wearable devices combined with outcomes data from electronic health records machines will be able to find patterns of diseases that we have been unable to perceive thus far.  Thus we may (hypothetically) find with the help of machines that someone whose HR goes above a certain value when exercising the day following a poor night’s sleep is more likely to present to the ED with cardiac symptoms.  

The next few years will see an explosion in the growth of wearable and invisible devices in healthcare and development of machine learning expertise to help interpret this data.  The question is whether physicians will jump in and help develop and research in this field or watch from the sidelines.