By adapting techniques used in weather forecasting, scientists were able to predict the timing of last year's flu season up to more than two months before its peak.
The team of scientists at Columbia University's Mailman School of Public Policy carried out their study in 108 cities across the country. Their findings were published Tuesday in the journal Nature Communications.
"Having greater advance warning of the timing and intensity of influenza outbreaks could prevent a portion of these influenza infections by providing actionable information to officials and the general public," said lead author Jeffrey Shaman, in a statement.
The flu season varies from year to year, stretching anywhere from October to May, so having knowledge of when the peak strikes could be would be useful for patients who want to know when to get their vaccinations, and for medical professionals who need to decide when and how many vaccines and antiviral drugs to stock.
The study builds on a previous study the team conducted in 2012, in which they retrospectively predicted the peaks of various flu seasons in New York City from 2003 to 2008.
But their new flu forecasting system combined data from Google Flu Trends, which estimates the number of outbreaks in an area based on the number of flu-related searchers, and region reports from the Centers for Disease Control and Prevention. Four weeks into the 2012-13 flu season, around the end of 2012, the system accurately predicted peaks in 63 percent of the country. Although the average peak-predictions were two to four weeks in advance, in some areas, the system predicted the peak up to nine weeks in advance.
Shaman said in the statement that the forecasting model "greatly outperformed" other approaches, which typically rely on using historical data to predict flu season peaks.
Still, the accuracy varied throughout different areas. The scientists were able to make more accurate predictions in smaller cities, for example. This could suggest that in bigger cities like New York City and Los Angeles, scientists may need to predict flu season peaks at a more granular level, such as by neighborhood.
The team plans to put the system back into action as soon as this year's flu season kicks in, and will publish its forecasts on a university website that will launch in the coming weeks.
Although there have already been a few reported cases of the flu this year, Shaman said the team will start making predictions "as soon as the needle starts to move."