Fitbit user data could be key to swifter population flu tracking

By combining prior CDC data with heart rate and sleep measurements, Scripps developed a new disease tracking model that they say could deliver speedy and accurate estimates.
By Dave Muoio
03:41 pm

Resting heart rate and sleep duration data collected from Fitbit devices could help inform timely and accurate models of population-level influenza trends, according to a new Scripps Research Translational Institute study published online in The Lancet Digital Health last Friday.

“Currently, CDC [influenza-like illness (ILI)] data are typically reported [one to three] weeks late and reported numbers are often revised months later,” the researchers wrote. “The ability to harness wearable device data at a large scale might help to improve objective, real-time estimates of ILI rates at a more local level, giving public health responders the ability to act quickly and precisely on suspected outbreaks."


By combining the readings of 47,249 users with three-week lagged ILI incidence rates from the CDC, researchers improved the Pearson correlation of their models across five states by as much as 32.9%, for an overall average of 0.12 compared to baseline. The prediction values of the updated models were strongly correlated with the agency’s final reported rates, yielding r values ranging from 0.84 (New York) to 0.97 (California).

Of note, the researchers also stressed that the values of their models did not change until the week of or week following similar action in the ILI rate, suggesting that the models aren’t so much forecasting future events as they are reflecting changes that are actually occurring.

“To our knowledge, this is the first study to evaluate and show that objective data collected from wearables significantly improved nowcasting of influenza-like illness,” the researchers wrote.


The Scripps researchers obtained and used de-identified sensor data from more than 200,000 US Fitbit users between the dates of March 1, 2016 to March 1, 2018. By honing in on users from five states who wore their device for at least 60 days and met other eligibility criteria, they landed on a final dataset of more than 13.3 million resting heart rate and sleep duration measurements.

The team set weekly thresholds for increased resting heart rates and sleep level that would denote periods of illness, and developed models that took these biomarkers into account alongside three-week lagged CDC ILI incidence data.


For years, technologists have been looking for ways their reams of data could support public health officials. Among the earlier and better known attempts was Google Flu Trends, a service launched in 2008 that used Google search trends as an early indicator of potential illness, but ultimately proved inconsistent and was shut down in 2015. Similar efforts in the years following have hinged on data from social media sites such as Twitter and Facebook, the latter of which released three new tools just last year.

There have also been efforts similar to the Scripps study that pull live updates from connected devices — for instance, Kinsa’s smart thermometer and insights platform.


“This study shows that using RHR and other metrics from wearables has the potential to improve realtime ILI surveillance,” the researchers wrote. “New wearables that include continuous sensors for temperature, blood pressure, pulse oximetry, ECG, or even cough recognition31,32 are likely to further improve our ability to identify population and even individual-level influenza activity. In the future, with access to real-time data from these devices, it might be possible to identify ILI rates on a daily, instead of weekly, basis, providing even more timely surveillance. As these devices become more ubiquitous, this sensor-based surveillance technique could even be applied at a more global level where surveillance sites and laboratories are not always available.”



The latest news in digital health delivered daily to your inbox.

Thank you for subscribing!
Error! Something went wrong!