A girl in front of a negative COVID-19 test walks upright while a boy in front of a positive COVID-19 test lays down with his eyes closed. In the center, a smartwatch screen features the time and a heart rate measure.
Design by Jennie Vang

Consumer-grade wearable devices, such as smartwatches, can monitor COVID-19 progression remotely through an analysis of heart rate data, according to a new University of Michigan study published on April 19. 

The study breaks down heart rate data into different physiological features, including basal heart rate — also referred to as resting heart rate — and heart rate response to physical activity. The study also examines the circadian rhythm in heart rate by analyzing a circadian amplitude parameter, or the heart rate variation over the day, and a circadian phase parameter, or the predicted time heart rate is at a minimum during the 24-hour circadian cycle. Another parameter is the autocorrelation parameter, which refers to the changes in heart rate due to mechanisms other than circadian variation, such as hormone release. These parameters change around COVID-19 symptom onset, which can help distinguish the infectious stage from healthy periods. 

According to the study, basal heart rate increases at symptom onset and the day after, which researchers hypothesized could be due to fever or heightened stress and anxiety, and generally starts to decline a few days after. The study found that heart rate response to physical activity increases soon after the symptom onset, and it increases more in individuals who experience coughing. 

Caleb Mayer, a doctoral student in the Mathematics Department and a co-author of this study, said the research is unique in the way it breaks down the single heart rate into several physiological systems.

“A lot of other work looks at heart rate as one signal and one system,” Mayer said. “And we’re really able to break it down into these different physiological systems and see how those differentially change, and they change in different time periods with respect to symptom onset.” 

The study’s analysis focuses on a “baseline” period, defined as anywhere from 8 to 35 days before COVID-19 symptom onset, and an “analysis” period, defined as anywhere from 7 days before symptom onset to 14 days after. 

“All of this analysis happens on a personalized, individual level because we know different individuals respond differently,” Mayer said. “They have different physiological systems. We wanted to break (it) down into a baseline period to get a sense of their normal balance of these parameters, and then an analysis period where we look more around symptoms when the disease starts affecting your physiology.”

The study analyzed participants from three separate studies that collected Fitbit and COVID-19 data: the Intern Health Study, which follows physicians in their first year of post-medical school training; a study by Tejaswini Mishra and colleagues, which collected smartwatch data from 32 individuals infected with COVID-19; and the Roadmap college student (Roadmap-CS) study, which collects data from undergraduate and graduate students at the University. 

Srijan Sen, director of the Frances and Kenneth Eisenberg and Family Depression Center and lead researcher of the Intern Health Study, said the data physicians typically work with is based on patients’ personal recollection of their symptoms. Wearable device data, on the other hand, provides real-time and objective information about disease progression, according to Sen.

“The hope is that (wearable devices) could really help us really understand what’s going on and come up with better tracking and treatments for a whole range of diseases,” Sen said.

According to Sen, while a direct clinical application of the research is not clear yet, this study represents progress in tracking the disease individually and in a larger population. 

“(The study) is advanced in that it gets us much closer to figuring out the right combination of these wearable signals to track for progression of the disease,” Sen said. “We’re not quite at applying it to individual patients yet, but hopefully we’ll progress both to getting more information on how this can be useful for individual patients and then broadly tracking the pandemic as it goes through different phases in local communities.”

Sen said he is excited to see how these wearable devices and mathematical algorithms can be used to track other diseases moving forward. 

“I’m excited about things that aren’t even particularly related to (COVID-19),” Sen said. “This paper (is about) tracking things related to (COVID-19) infection, but I suspect it’s also tracking numerous other things like other stresses and other infections … Using these wearables (allows us) to understand a lot more about how the environment is affecting our lives.”

Sen said the study shows the value of wearable devices in helping patients improve their understanding of biological processes and the state of their health.

“I think (the study) really demonstrates the value of these wearable devices and providing information that I think can eventually be clinically useful,” Sen said. “It also lets people understand what’s going on in their bodies in different and unique ways, and hopefully helps people with healthier lives.”

Sung Won Choi, associate professor of pediatrics at Michigan Medicine and lead researcher of the Roadmap-CS study, said the study was initially developed for cancer patients to improve their well-being using a mobile app and wearable devices. When the pandemic hit, the Roadmap-CS researchers started to offer the mobile app and wearables to healthcare workers in May 2020. At the start of the following academic year, researchers decided to include U-M students in their research. 

“(U-M) students were about to come back to campus in August,” Choi said. “Wouldn’t that be amazing if we could … offer this as a potential tool for (students) to help support their wellness and well-being and also provide them with a wearable sensor like Fitbit that could potentially track the data? And that’s exactly what we did.”

Choi said the COVID-19 data from students, along with the other two datasets, enabled the application of the mathematical algorithm to the heart rate data.

“During that time, September to December 2020, that’s when we had this really strict quarantine isolation,” Choi said. “In order to use these heart rate algorithms for (COVID-19), you need a large number of positive cases. So that’s why it was important to combine all these datasets.”

Choi said incorporating remote monitoring techniques is beneficial to physicians and patients overall with the growth of telehealth, or remote healthcare services, during the pandemic. 

“The pandemic accelerated and made us do (telehealth),” Choi said. “I think more in-hospital care is moving to the outpatient setting. Being able to monitor patients non-invasively and remotely is a huge advantage.” 

Choi emphasized the importance of collaboration between physicians and data scientists in the medical field.

“I think there’s a lot of potential for these wearables and for data science, and I can’t express enough the importance of collaboration,” Choi said. “It goes beyond just depending on physicians. You need a team to be able to look at these types of problems.”

Daily Staff Reporter Jingqi Zhu can be reached at jingqiz@umich.edu.