American researchers have studied if Fitbit data can predict the beginning of a flu outbreak.
First study to determine impact of Fitbit data on flu prediction
Influenza (commonly called the flu) is a viral infection that impacts the respiratory system. It is a significant disease, resulting in 650,000 yearly deaths worldwide, including roughly7% of working age adults and 20% of children under five years of age.
Predicting when flu outbreaks will occur is a challenge, as traditional influenza surveillance methods have a reporting lag of 1-3 weeks. This lag can allow outbreaks to proceed unnoticed, and to spread. Opportunities are thus lost for healthcare authorities to react quickly to halt outbreaks by ensuring that people with the flu stay at home, get appropriate medications, and follow good hygiene practices such as hand washing.
One technology that has never been investigated for use in influenza surveillance is that of wearable devices such as Fitbit. To this end, researchers in the United States conducted the first-ever study to see if wearable devices can aid in real-time monitoring of influenza activity.
Heartbeat and sleep data from 47,248 Fitbit users
The study used de-identified Fitbit data from a population of Fitbit users, summarized as follows:
Number of Fitbit users: 47,248
Locations: California, Texas, New York, Illinois, and Pennsylvania.
Average age: 43 years old.
Male/Female split: 40%/60%.
Fitbit usage: wearing of one specific unique Fitbit device for at least two months during the period March 2016-March 2018.
After collecting these data (over 13 million individual measurements), the researchers calculated resting heartbeats and sleep durations. They then categorized users as abnormal if their average weekly resting heartbeat was above their overall weekly average, or if their average weekly sleep duration was below their overall weekly average.
The researchers then compared this calculated Fitbit data to weekly estimations of influenza occurrences from the American Centers for Disease Control (CDC).
Flu prediction improved
For all five states covered in this study, there was a measurable statistical improvement in real-time monitoring. As well, the researchers were able to create a statistical definition of what a flu suffer would look like, based on alignment with CDC data.
Study limited by amount of Fitbit data
Some of the study limitations noted by the researchers included:
- An overall lack of Fitbit data. This lack prevented the researchers from accounting for factors that affect activity, such as seasonal changes and illness.
- The fact that resting heart rate can be affected by many factors, such as stress, and not just by the flu.
- That sleep measurement devices have been shown to have low accuracy, although their accuracy has been improving as new technology comes onto the market.
The results of this study suggest that Fitbit and other wearable devices have the potential to aid in the prediction and response to seasonal flu cycles. This prediction may become even more accurate as more data from these devices become available, and as the sensing accuracy of the technology increases.
Written by Raymond Quan
Radin et al. “Harnessing wearable device data to improve state-level real-time surveillance of influenza-like illness in the USA: a population-based study.” Lancet Digital Health 2020 https://doi.org/10.1016/S2589-7500(19)30222-5
Image by StockSnap from Pixabay