Your smartwatch could tell if you have an infection days before you develop a fever

Tuesday, June 2, 2020
By Lillian Clark
Image depicting smartwatch and viruses

News of the first U.S. coronavirus case hit the American public on Tuesday, Jan. 21. For Jessilyn Dunn, assistant professor of biomedical engineering at Duke University, the infection in Washington State was a cue to launch a new study⁠—CovIdentify

Duke CovIdentify researchers are hoping to use smartwatches to catch early signs of COVID-19 infection and maybe even predict how severe patients’ symptoms will be. Health measurements via smartwatches or other wearable devices, such as heart rate, sleep, and activity, have helped scientists detect flu-like illnesses before. The Duke Center for Applied Genomics and Precision Medicine  (CAGPM) has partnered with Jessilyn Dunn on this study.

CovIdentify isn’t so different from Dunn’s pre-coronavirus work. Before coming to Duke, she worked with wearable devices for infectious disease detection at Stanford University. She realized almost immediately her research had applications in the pandemic.

“With our algorithms that we developed, we could actually tell that people were sick usually between 72 to 48 hours before they reported sick. There were some obvious signs on their smartwatches,” Dunn said of her Stanford research.

Studies at Scripps Translational Research Institute, UCSF, and Stanford Medicine’s Healthcare Innovation Lab are also tackling the pandemic with wearable data. UCSF’s TemPredict study has given Oura smart rings to 2,000 front-line health workers. The Scripps study, called DETECT (Digital Engagement & Tracking for Early Control & Treatment), hopes to provide an early-warning map of viral outbreaks.

Brinnae Bent, doctoral candidate in biomedical engineering working on the CovIdentify team, mentioned CovIdentify’s international scale sets it apart. Their website is available in Mandarin, Korean, Spanish, and Hindi.

Jennifer Radin, senior scientist at Scripps and lead author on DETECT, published a paper in January in which she used data from 47,000 Fitbit-wearers to improve CDC influenza predictions by as much as 32%.

Even though many wearables don’t measure skin temperature, Radin explained resting heart rate provides early insights. “Resting heart rate actually increases before temperature increases and before symptoms increase,” Radin said.

Unfortunately, people can’t just take their pulse as a DIY infection test. These studies draw conclusions from slight variations after first finding a baseline for each individual’s ‘normal.’ Mark Wanner, science writer and Fitbit wearer, believes it takes several months to establish a decent heart rate baseline and a full year of ups and downs to have a strong one. 

“I have noticed that when I’m coming down with an illness, my resting heart rate will increase by several bpm, usually starting just before symptoms appear, and it will stay elevated until I am nearly recovered,” Wanner said. Wanner also sees an increased resting heart rate if he drinks more alcohol than usual.

It will be challenging to differentiate between infections with sensor data alone⁠—changes in activity and heart rate could point to the flu as much as they could coronavirus. That’s why DETECT and CovIdentify have both incorporated a survey element where participants can report symptoms and COVID-19 test results. Tagging data could help these researchers notice signs which identify the infection a participant has.

These studies are slated to take at least two years, but Radin hopes DETECT will see results soon. “We’re already getting people reporting that they have different symptoms, and even a few people have reported coronavirus positive tests, so we can start looking at these individuals and seeing when their data started deviating from norm and start improving our hypotheses,” she said.

Wearable users tend to be wealthier than the general population. Lately Dunn spends her time racing for funding to take the edge off this socioeconomic skew.

“There are a lot of people in underserved populations that are particularly hard hit by COVID, who typically would not buy a wearable or may not have the money to afford a wearable,” Dunn said. She wants to avoid excluding them from the study and contributing to existing disparities. She continued, “We’re trying very hard to secure funding so that we can be able to purchase devices for people who are mostly essential workers, who are grocery store workers, delivery people, and hospital cleaning staff that would otherwise not have devices,” Dunn said.

CovIdentify currently has a 20% device discount available with Garmin and Dunn said discounts for Fitbit devices are coming soon.

Dr. Geoffrey Ginsburg, director of CAGPM and co-lead on CovIdentify, sees the recent move towards telemedicine due to coronavirus as a paradigm shift that could make healthcare and research more accessible to everyone through technologies like wearables.

“A lot of patients don’t like coming to hospitals because they have trouble parking or they can’t take time away from work or it’s just too costly for them to see a doctor rather than to stay at work or to take care of their child. Now we’ve created a platform that eliminates the commute, eliminates the parking hassles, and it can make for a very focused 30 minute visit,” Ginsburg said.

As digital medicine becomes more important, Bent believes developing standards for verifying the accuracy of these devices will be essential.

A 2019 Johns Hopkins study stated that no reliable mechanism existed to find which digital health technologies were actually validated. According to Statista, there were over 44,000 mobile health apps available in the Apple store in the third quarter of 2019.

“I hate to use this overused phrase of the ‘Wild West,’ but that’s kind of the way the digital health space is right now. Anybody can build something and market it to do something without ever having to prove that it actually can do what it’s meant to,” Ginsburg said. He said wearable validation has improved slightly as some devices have gone through the FDA approval process.

Wearables take hundreds of thousands of measurements per day, whereas doctor’s visits are infrequent. Constant monitoring is the basis of precision medicine.

Dr. Ginsburg said, “Up until the last five or six years, I would say that medicine has been practiced by people coming to the hospital and getting data at one point in time, and then they go away and you don’t hear from them until they come back. Now we have an opportunity in medicine to effectively get data all the time on our patients.”

James Thomas, associate professor of epidemiology at UNC-Chapel Hill, believes the trade off with digital devices has always been between convenience and privacy. 

Although data from your smartwatch is private and belongs to companies like Apple and Google he said, he warned that some governments may choose to sidestep regulations and buy data for public health surveillance.

“Twenty-five, fifty years ago, the M.O. was to collect as little data as needed, because it was so expensive. Now, because it’s so cheap, the M.O. is to collect as much data as possible. And there are indeed some very good uses of data. They’re very powerful and they will allow us to do things to control an epidemic or a pandemic. That’s the case, until it is misused for some other purpose,” Thomas said.

Thomas spends his time advising North Carolina on its coronavirus response and has created a pandemic ethics site to help policymakers with tough decisions.

Dunn was careful to emphasize that CovIdentify is not a contact tracing study. Participants can provide their zip code and address in the survey, but that’s optional, and the researchers won’t collect any other location information. They are making every effort to keep participant data unbreachable on secure Duke servers.

These days, Dunn hosts virtual coffee chats with her lab, spending 15 minutes each day to stay connected before they split off and work. They have accomplished in less than three months what normally could have taken six or more.

A thousand participants joined CovIdentify in its first week, but ultimately Dunn hopes to gather data from hundreds of thousands of people.

“The success will be really dependent on the outreach,” Ginsburg said.

Original story published on MediaHub on May 29, 2020. 

More Coverage

'Covidentify' Adapts to Address a Changing Pandemic

How Your Smartwatch May Screen for COVID19

Wearable tech can spot coronavirus symptoms before you even realize you’re sick

Duke Researchers Look to Wearable Technology to Detect Coronavirus

Can wearable health devices help detect COVID-19?

Duke researchers to collect biometrics from wearables in search for early COVID-19 detection method

Can your Garmin detect coronavirus?

CovIdentify Study Aims to Monitor COVID-19 Spread Through Wearable Smart Devices

Duke researchers say smartphones, data may be key weapon in fight against COVID-19

‘CovIdentify’ Pits Smartphones and Wearable Tech Against the Coronavirus