Skip to main content

Analysis

A Novel Surveillance Tool for Flu: Fitbit

By MedPage Today  
   January 17, 2020

Heart rate, activity, and sleep data hold promise for population-level influenza tracking.

This article was first published on Thursday, January 16, 2020 in MedPage Today.

By Molly Walker, Associate Editor, MedPage Today.

Fitbits, wearable devices that measure resting heart rate and sleep time, hold promise in measuring flu at the state level, researchers found.

Weeks during which deidentified Fitbit users in five states had elevated heart rates and more sleep time tended to be those when influenza-like illnesses (ILI) were most common in those states, reported Jennifer Radin, PhD, of Scripps Research in La Jolla, California, and colleagues, writing in The Lancet Digital Health.

When the Fitbit data were included in flu-intensity prediction models, Pearson correlations increased by an average 0.12 points (SD 0.07) over the original models, they said, adding, "Correlations of the final models with the CDC ILI rates ranged from 0.84 to 0.97."

"Responding more quickly to influenza outbreaks can prevent further spread and infection, and we were curious to see if sensor data could improve real-time surveillance at the state level," Radin said in a statement.

Google Flu Trends and social media tools like Twitter have attempted to capture real-time flu surveillance data, but the authors noted they have been less than successful, with Google Flu Trends missing early waves of the 2009 H1N1 pandemic strain, as well as overestimating activity during flu outbreaks. They lamented the lack of "objective data streams" to provide real-time flu activity information.

Enter wearable sensors. The authors hypothesized that fitness bands or smart watches "might be able to identify abnormal fluctuations indicting perturbations in one's health, such as an acute infection." They pointed out that acute infections often raise the resting heart rate, and sleep and activity patterns differ when someone is not feeling well.

Radin and colleagues obtained deidentified Fitbit data from a convenience sample of users from March 2016 to March 2018. Users wore a Fitbit for at least 60 days during the study and had only one Fitbit the entire time. The authors noted that to measure population-level changes, they included data from the five states with the most Fitbit use: California, Texas, New York, Illinois, and Pennsylvania.

The Fitbit data were then correlated with ILI reports from the CDC, which generates weekly estimates at the state level.

Overall, about 47,000 users had data included. Users were a mean age of around 43 and 60% were women. Fitbit data improved flu predictions in all five states, the researchers said, citing an improvement of 6% to 33% over baseline models.

"To our knowledge, this is the first study to evaluate the use of [resting heart rate] and sleep data in a large population to predict real-time [influenza-like illness] rates at the state level," the authors wrote.

An accompanying editorial by Cecile Viboud, PhD, of the NIH, and Mauricio Santillana, PhD, of Boston Children's Hospital, characterized the study as "a promising first step towards integrating wearable device measurements in predictive models of infectious diseases."

"We anticipate that Fitbit data could be used as one of several external covariates in predictive models for influenza, along with other health, digital, and social media indicators," the editorialists wrote. "Analysis of repeated individual biological measurements, such as those provided by Fitbit devices, is an enticing way to monitor population health, because measurements are passive, high volume, and noninvasive."

However, they noted that due to interannual variability in flu, additional data would likely require a longer-term research agreement with Fitbit.

Limitations to the data include a lack of an activity variable, which the authors said could control for seasonal fitness or short-term activity changes. Weekly resting heart rate averages might incorporate data where a person is both sick and not sick, which could underestimate illness by lowering the weekly averages, they noted. They also cited sleep measuring devices' low accuracy, though they said it is improving.

"In the future as these devices improve, and with access to 24/7 real-time data, it may be possible to identify rates of influenza on a daily instead of weekly basis," Radin said.

This study was supported by the NIH National Center for Advancing Translational Health.

The authors disclosed no conflicts of interest.

Viboud and Santillana disclosed no conflicts of interest.

“We were curious to see if sensor data could improve real-time surveillance at the state level.”


KEY TAKEAWAYS

Google Flu Trends and social media tools like Twitter have attempted to capture real-time flu surveillance data, but they have been less than successful.

Researchers hypothesize that fitness bands 'might be able to identify abnormal fluctuations indicting perturbations in one's health, such as an acute infection.'

They pointed out that acute infections often raise the resting heart rate, and sleep and activity patterns differ when someone is not feeling well.


Get the latest on healthcare leadership in your inbox.