A new statistical model, built on search data from Google and principles from weather modeling, can predict flu outbreaks up to seven weeks in advance.
The results, published Monday in the Proceedings of the National Academy of Sciences, signify a transition in the study of infectious disease from modeling past outbreaks and events to predicting future ones.
The researchers, Jeffrey Shaman of Columbia University and Alicia Karspeck of the National Center for Atmospheric Research in Boulder, Colo., used data from the Google Flu Trends project, which keeps track of searches for flu-related topics and ties them to the geographic location of the searcher. Such data is now available for 28 countries as well as many local areas within those countries. The study focused on New York City.
The Google project made waves in 2008 when its results nearly matched those of the U.S. Centers for Disease Control and Prevention. The researchers’ model takes into account the importance of current conditions in projecting the timing of a flu outbreak.