Model Predicts Flu Season Peak Seven Weeks in Advance

(HealthDay News) – Using a technique commonly applied in weather prediction, peaks of influenza season can be predicted seven weeks in advance, according to a study published online Nov. 26 in the Proceedings of the National Academy of Sciences.

Jeffrey Shaman, PhD, from Columbia University in New York City, and Alicia Karspeck, PhD, from the National Center for Atmospheric Research in Boulder, CO, used a data assimilation technique commonly applied in numerical weather prediction to initiate real–time forecasts of seasonal influenza outbreaks. Real–time, Web-based estimates of local influenza rates (Google Flu Trends) were used for this forecasting.

The researchers found that, following assimilation of these estimates for the 2003–2008 influenza seasons in New York City, retrospective ensemble forecasts were generated on a weekly basis and real–time predictions of peak timing were made more than seven weeks in advance of the actual peak. Based on the spread of the forecast ensemble, confidence in these predictions could be inferred.

“This work represents an initial step in the development of a statistically rigorous system for real–time forecast of seasonal influenza,” the authors write.

Full Text (subscription or payment may be required)