Researchers have used data gathered by connected thermometers to predict the onset of an influenza outbreak. An analysis of data from users of the iThermometer wearable and its companion app in China predicted the spike in the 2015-16 flu season one month before the country’s public health body.
A team working out of Boston, Massachusetts, published the analysis in the American Journal of Public Health. The study looked at 44,999 observations gathered by online health educational tool Thermia from parents in China.
Almost all the observations were generated by iThermometer, a wearable aimed at parents who want to keep track of the temperature of their children. The system prompts users to provide additional information on the symptoms of their children when readings are taken.
Retrospectively sifting through this data showed the resource could predict the onset of the 2016 flu season one month before China’s National Health and Family Planning Commission. The public health body relies on reports from healthcare professionals to generate its assessments of the flu season, an approach that suffers from a time lag and inconsistencies in classification.
A team working out of Boston Children’s Hospital thinks connected thermometers such as Raiing Medical’s FDA-cleared iThermometer can improve on this approach, particularly in China.
“The fact that we were able to predict influenza outbreaks faster than China's national surveillance programs really shows the capacity for everyday, wearable digital health devices to track the spread of disease at the population level,” lead study author Yulin Hswen said in a statement. "In geographically large and densely populated countries like China, tools like Thermia can provide better on-the-ground disease surveillance than by relying on data that is only captured at the point of treatment in the clinic.”
Precedent suggests Thermia must successfully make predictions around the world during multiple flu seasons, rather than retrospectively, before it can convincingly claim to be better than current data gathering techniques.
Google published a paper in Nature in 2008 showing how its search data could deliver near-instant predictions of flu trends. Over the next few years, it continued to predict outbreaks before the U.S. Centers for Disease Control and Prevention. But the project went off the rails in 2011, culminating in the tool missing the peak of the 2013 flu season by a wide margin. Google subsequently killed off the flu project, although other groups continue to explore the predictive power of online data.
The Thermia project differs from Google’s program and others that use social media data, however. It has access to patient-reported outcomes and temperature readings from connected thermometers, giving researchers access to data more definitively tied to health than the flu-related search terms Google relied on. But the reliability of the dataset could be compromised by repeat entries from the same parent.