Could Your Fitbit Help Detect the Flu?

By | January 21, 2020

Jan. 17, 2020 — Wearable activity monitors that also measure your heart rate could one day warn people they might be getting sick with an infection like the flu.

Doctors have long known that a higher resting heart rate — the number of times each minute the heart beats while a person is sitting or sleeping — can be a sign that the body’s immune system is ramping up for a fight.

For example, research has shown that young men with fevers had increases in their resting heart rate of about 8.5 beats per minute (bpm) for about every 2-degree Fahrenheit increase in body temperature. Other studies have shown that a child’s resting heart rate can go up even more — between 10 and 14 bpm for every 2-degree Fahrenheit rise in temperature.

But there’s a wide variation in what’s normal from person to person, and doctors may not be able to pick up on it.

“When a doctor sees you, as long as your heart rate is between 60 and 100, we don’t think about it. That’s within the normal population range,” says study author Steven Steinhubl, MD, a cardiologist and director of digital medicine at Scripps Research Translational Institute in La Jolla, CA.

“But if we knew that my resting heart rate every day was typically 60, and I come to the doctor’s office and my resting heart rate is now 68 or 72. If we had the knowledge to say, ‘Hmmm. That’s unusual.’ Maybe that’s a sign that something is going on. But we’ve never really had that before,” he says.

Researchers at Scripps wondered if they could take advantage of the Fitbit’s ability to track both heart rate and activity to learn when large groups of people were getting sick — for example — during flu season. Their study is published in The Lancet Digital Health.

A note of disclosure: Eric Topol, MD, who is editor-in-chief of WebMD’s sister site, Medscape, was one of those study authors, but he was not interviewed for this story. The study was partly funded by a grant from the National Institutes of Health.

See also  Hardcore Self Help: F**k Anxiety (Volume 1)

How They Did It

The researchers approached Fitbit with their idea, and the company gave them access to 2 years of its proprietary data for 200,000 Fitbit users from 2016 to 2018. They focused on five states that had the greatest numbers of users: California, Texas, New York, Illinois, and Pennsylvania. The data was stripped of personal identifying information so it couldn’t be traced back to a particular person. There’s a clause in Fitbit’s privacy policy that says users’ de-identified data may be used for research.

Study authors eliminated users who had missing data — meaning they’d taken their Fitbits off during the day for some reason — and those who didn’t wear their devices for at least 1,000 minutes, or about 16 hours, each day. They were left with data from about 47,000 steady Fitbit users. On average, these regular users were around 43 years old, and 60% were women.

The Fitbit takes a resting heart rate reading when a person has been motionless for a set amount of time — for example, 5 minutes. Scientists compared those readings with sleep data. They expected that someone with a higher-than-average resting heart rate, who was also sleeping more than usual, might be coming down with the flu.

They compared their Fitbit-measured “flu” cases to weekly cases of influenza-like illness reported by the CDC for each state across two flu seasons. They were a pretty close match.

It turns out the mathematical model built by the scientists could detect higher flu activity about the same time it was happening on the ground, or shortly after. But it wasn’t as good at giving an early warning that an outbreak might be coming. The researchers think that might be because infections like the flu raise a person’s heart rate for several days or even weeks.

See also  New blood test could be used to help millions infected with TB

The Fitbit warning system isn’t perfect. It can tell when something is off with a person’s resting heart rate, but it can’t really tell why. You would still need to touch base with a doctor or other health care provider to tell if it’s caused by an infection or something else. Not getting enough sleep can raise your resting heart rate, for example. Stress can, too.

Still, the study shows that fitness trackers can help tease out telltale signs of illness.

Previous studies have looked at whether the Apple Watch’s heart rate tracker may warn people of irregular heart rates, the kind that happen with atrial fibrillation. The jury is still out about whether that function does more good than harm.

Steinhubl says doctors don’t know exactly when your heart rate goes up during an infection. There is some evidence that an increase in heart rate might be one of the first signs that someone is getting sick. Studies of Ebola infections in primates, for example, found that changes in heart rate happened about 48 hours before the start of a fever. Another study, he says, in children with asthma, found that one of the strongest predictors of an attack was a higher heart rate the night before.

Steinhubl says getting an early warning from your activity tracker or smartwatch that something is coming might give people more time to take advantage of antiviral medications like oseltamivir (Tamiflu), which help only if you take them in the early days of an infection.

The ability to do that for individuals is still far off in the future.

‘Very Intriguing’

For now, though, an expert in disease forecasting who was not involved in the study says Fitbit data seems to be a new and interesting way to track the flu.

“The really interesting thing is that it’s using a new source of data that’s never been used before to track infectious disease outbreaks,” says Cecile Viboud, PhD, a mathematical epidemiologist at the Fogarty International Center at the National Institutes of Health. Viboud wrote a comment on the study.

See also  Using smart watches to monitor your heart could do more harm than good

“It’s also very intriguing that something very nonspecific, like a change in resting heart rate, you can sort of pinpoint influenza epidemics on the population level,” she says.

Viboud points out that the study covered only two flu seasons, and each flu season is different, so it’s not clear if the results would hold over a longer period of time.

She says flu forecasters know that models that rely on many different sources of data, like search data from Google, along with data from social media sites and electronic medical records, are a bit better at predicting flu than the Fitbit data that was modeled for this study.

She says she hopes data like this can be combined with other sources to give the health care system more warning when the flu might be about to hit a certain part of the country.

Right now, Viboud says, flu forecasting models can predict where infections will spread about 3-4 weeks in advance.

“If we could do a little better by using even more real-time data, maybe get to about a month or a month and a half, that would be really useful,” she says.

That would give doctors more time to urge their patient to get flu shots. Local pharmacies could stock more antiviral medications. And hospitals could reschedule surgeries if they knew they were going to get a surge of patients.

Sources

Steven Steinhubl, MD, director of digital medicine, Scripps Research Translational Institute, La Jolla, CA.

Cecile Viboud, PhD, mathematical epidemiologist, Fogarty International Center, National Institutes of Health.

The Lancet Digital Health: “Fitbit-informed influenza forecasts.”

STAT: “First Opinion,” March 15, 2019.

© 2020 WebMD, LLC. All rights reserved.

WebMD Health