Detecting coronavirus in real-time

Researchers at Stanford University’s Department of Genetics are harnessing the treasure trove of data that is being recorded 24/7 by smartwatches. Although wearable tech, including rings, wristbands and even sensors that can be incorporated into clothing, can do more than just measure heart rate and daily steps, it would appear that this simple data alone can provide powerful insight into the real-time health status of its wearer.

By using a personalised disease detection algorithm, the MyPHD app, developed by Mike Snyder and his research team, uses the data from smartwatches to detect what they call “stress events”. These stress events can include poor sleep, mental health stress, vaccination and even respiratory viral infections.

In fact, the algorithm was able to detect 80% of coronavirus cases, through which red alerts were generated on average 3 days before the onset of any symptoms. The algorithm even produced alerts for 14 out of 18 asymptomatic cases, on average 5.5 days before diagnosis. Mike believes that this real-time alerting system could be a powerful technology in the toolbox to help fight the coronavirus pandemic. “We can find people many days in advance of symptoms, or even if they have no symptoms at all, and by alerting them, ultimately what we hope to be able to do is have them self isolate, or get tested early”.

So, how does the algorithm work? Mike explains: “It will follow your healthy baseline, and when you have a jump up in resting heart rate, it will send off an alert. [...] We don’t only use resting heart rate, we do add a little bit of step and sleep data [as well]”. The app is easy to interpret due to a nifty traffic light system. A green light means there’s nothing out of the ordinary, whereas a red light means something has changed, and triggers an alert. “If you’re just doing your normal thing, and a red alert goes off, something is up. You’re either having a mental health situation, or perhaps you're ill from a viral infection”.

Currently the wearer has to contextualise the red alert by responding to questionnaires about lifestyle and health, but Mike hopes that in the future, they will be able to filter out the alerts from some normal activities, by using data such as heart rate variability, and respiration rate - the number of breaths a person takes per minute. “We think that mental health will probably have a more direct effect on heart rate, whereas respiratory viruses both affect heart rate as well as respiration, so you can actually start teasing out these different signals. [...] The more data we can bring in, I know the more specific we’ll be in trying to do these sorts of diagnoses”.

Where does the future lie for wearable technology? A quick bit of online shopping reveals that more and more sensors are being added to these devices. Measuring body temperature, the level of oxygenation in your blood, and even the conductance of your skin. Mike thinks this is just the beginning. “Ultimately there’ll be things like implantables, that you put under your skin, that measure you passively, to be able to follow your health at a detail that we don’t currently do”.