Google is shuttering its flu tracking program known as Flu Trends and will funnel the search data powering it to public health researchers instead.
Data scientists who are deploying algorithms to monitor diseases at Boston Children’s Hospital, Columbia University’s Mailman School of Public Health, and the Centers for Disease Control and Prevention will receive the Google flu and dengue data, Google said in a blog post Thursday.
The HealthMap group at Boston Children’s and Harvard Medical School uses public information on the Web like tweets, news reports, and data on Google Trends to track the emergence and incidence of disease outbreaks world-over, including monitoring Ebola outbreaks. The idea is that such tools will help public health agencies stay ahead of outbreaks, and help them make decisions.
Google has been collecting and displaying search data such as the location, frequency, and terms searched for related to flu, in the hopes of alerting public health officials of outbreaks. While Google will no longer display the data itself, the HealthMap group will have direct access to the Google search data. “It will improve our models significantly,” he said, said John Brownstein, leader of the research group and chief innovation officer at Boston Children’s.
Now there are dozens of researchers the world over who are fine-tuning models to monitor and predict disease outbreaks using statistics, modeling, and data mining.
When it launched in 2008, Google Flu Trends was among the first to display that information on a large scale. But it wasn’t always right — in 2012, the system overestimated the US flu incidence, predicting a season twice as bad as the one that hit.
What went awry? Brownstein said that “in a constantly evolving field” predictive models must stay on their toes to keep up with changes in digital behavior. The frequency with which people search for medical terms changes, for example, as could the reasons motivating their search.
Still, it was an effective pioneer. “It was a success in bringing attention to new data sources and the idea of digital data to public health,” Brownstein said.