University of Iowa researchers were able to predict flu activity weeks ahead of the CDC with help from the Kinsa smart thermometer, shown above. PHOTO UI
By Katharine Carlon
katharine@corridorbusiness.com
This year’s flu is already one of the worst in recent memory, responsible for sky-high absenteeism, packed clinic offices and nearly 100 Iowa deaths as of last week.
And almost no one saw it coming.
Almost no one, that is, except University of Iowa researchers Aaron Miller and Dr. Philip Polgreen, who called it months ago when their analysis of data from a “smart thermometer” phone app suggested 2018 would bring a particularly rough flu season.
“While our study was still under review, we kept looking at the data and realized this would be a much worse flu season than the last few years,” said Mr. Miller, a UI postdoctoral researcher in computer science who led a just-published study in the journal Clinical Infectious Diseases on tracking influenza via smartphone app. “Once our study was accepted, we updated it, and were able to show we predicted weeks before the [Centers for Disease Control and Prevention] that this was going to be a very bad year.”
The CDC relies on aggregated reports from around the country on clinic visits, hospitalizations and flu-related deaths, Mr. Miller said, making its data, at best, a couple of weeks behind real-time flu activity.
By contrast, the UI study analyzed the results of 8 million temperature readings from 450,000 smartphone thermometer devices nationwide and found it was possible to not only take up-to-the-minute snapshots of national and regional flu activity, but to forecast influenza levels two to three weeks into the future.
In short, researchers found their real-time results almost exactly mirrored what the CDC put out weeks later. That, they say, means estimates of future flu activity could be improved by a month or more.
“Our findings suggest that data from smart thermometers are a new source of information for accurately tracking influenza in advance of standard approaches,” said Dr. Polgreen, UI associate professor of internal medicine and epidemiology and the study’s senior author. He added that getting an earlier jump on the spread of flu could “help alert health care professionals that influenza is circulating, help coordinate response efforts, and help anticipate clinic and hospital staffing needs associated with high levels of influenza activity.”
To conduct the study, Mr. Miller and Dr. Polgreen sifted through anonymized data provided by Kinsa Inc., the company that makes and sells the Smart Ear Thermometer, looking for fever patterns that aligned with influenza. Kinsa, headed by former Clinton Health Access Initiative vice president Inder Singh, developed the ultra-low-cost smartphone-connected thermometer to help users track symptoms, get advice directly from medical professionals and get a sense of the “health weather” in their localities when infectious diseases like flu or strep throat are circulating.
But Mr. Singh, another co-author of the UI study, along with Kinsa Product Manager Erin Koehler, hopes his device – able to nearly instantly spot fever spikes in states, cities and even neighborhoods – will one day serve an even higher purpose, creating a real-time “map of human health” to track and stop the spread of communicable illness.
“Similar to advance notice of extreme weather, early signs of increasing illness could be invaluable to families and communities, enabling people to take precautions in areas that will be hard hit in the weeks to come,” Mr. Singh said in a release last month touting his company’s accurate and early prediction of the flu season.
“Kinsa’s mission is to stop infectious illness from spreading through early detection and early response, and this flu season is shaping up to be one where advanced knowledge could potentially save lives.”
The UI study is among the first to independently analyze Kinsa’s data, which Mr. Miller believes will be an important forecasting tool going forward. Other attempts at early data collection using Twitter and the now-abandoned Google Flu Trends tool were unsuccessful, missing the peak of the 2013-2014 flu season, for example.
“Those used proxy information from people searching ‘fever’ or their symptoms, words like ‘cough’ and ‘flu,’” Mr. Miller said. “This is the first time we’ve actually been able to measure symptoms in real-time as opposed to interest in symptoms.”
In addition to tracking the spread and geographic location of flu outbreak, researchers found the smart thermometer data also allowed them to estimate which age groups were being most affected during the season. They found, for example, that influenza-like illness moved more frequently from children to adults than the reverse.
Monitoring the duration of fever from the smart thermometer readings also revealed fevers occurring during flu season were most likely to last three to six days, and that instances where users had fever that went away for a few days and then returned in what’s known as a “biphasic” fever pattern often reflected a secondary bacterial infection such as flu-induced pneumonia.
“Being able to track the duration of fevers and the return of fevers could help us learn more about an influenza season as it is emerging,” Dr. Polgreen said, “and possibly provide information about other infections, including new emerging infectious diseases.”
One of the next areas the group is hoping to explore is the identification of other types of influenza, Mr. Miller said, but the technology could also be used for activities such as monitoring surgical site infections, “so that when you send patients home, you have a real time feedback loop.”
He added it may eventually be possible to track flu at a granular, school-by-school level. And as the smart technology evolves and becomes more widespread, data may become more of a two-way street.
“Because this data is being generated by users of a device, we have a platform for public health intervention,” Mr. Miller said. “We can’t do it yet, but at some point, we can feed back information through the device. Messages like ‘get a flu shot.’”
