New Flu Forecasting Model Uses Google Search Data
By combining data that keeps track of Google searches about flu with the latest techniques used in weather forecasting, two US researchers have developed a computer model that predicts regional peaks in flu outbreaks more than 7 weeks ahead. They hope their system will one day help health authorities and the general public better prepare for seasonal flu outbreaks.
Jeffrey Shaman, an assistant professor of Environmental Health Sciences at Columbia University's Mailman School of Public Health in New York, and Alicia Karspeck, a a climatologist with the National Center for Atmospheric Research in Boulder, Colorado, write about their new model in the 26 November online before print issue of the Proceedings of the National Academy of Sciences, PNAS.
The peak of seasonal flu outbreaks varies hugely from year to year and region to region. In the temperate regions of the Northern Hemisphere, outbreaks can peak any time between October and April.
Shaman says in a statement to the press that their model offers a "window into what can happen week to week as flu prevalence rises and falls".
He says he can see a future where such flu forecasts will run alongside weather reports on local TV news.
Like the weather, flu conditions vary from region to region, he says. For instance, Atlanta might peak weeks before Anchorage.
"Because we are all familiar with weather broadcasts, when we hear that there is a 80% chance of rain, we all have an intuitive sense of whether or not we should carry an umbrella," says Shaman.
"I expect we will develop a similar comfort level and confidence in flu forecasts and develop an intuition of what we should do to protect ourselves in response to different forecast outcomes," he adds.
This could prompt people to be more aware of how they feel and to take precautions, such as get a flu vaccine and be more careful around people sneezing and coughing, say the researchers.
On another level, such forecasts can help authorities plan stocks of vaccines and antivirals, and in the case of severe outbreaks, whether to close schools, for instance.
Shaman and Karspeck applied principles used in weather forecasting into their model. One of these, is to fine-tune it with real-time observational data, and thus "nudge the model to conform with reality and reduce error in the model simulations," explains Shaman.
To test their model, they used data related to the 2003-2008 influenza seasons in New York City.
From Google Flu Trends they obtained "near-real-time data" which estimates outbreaks based on flu-related searches in a given region. When they plugged the data into their model, and compared it to the actual patterns of the outbreaks, they found the model could predict the peaks of the outbreaks more than seven weeks ahead.
They conclude their work represents "an initial step in the development of a statistically rigorous system for real-time forecast of seasonal influenza".
Irene Eckstrand of the National Institute of General Medical Sciences at the National Institutes of Health (NIH), which helped fund the study, says:
"Flu forecasting has the potential to significantly improve our ability to prepare for and manage the seasonal flu outbreaks that strike each year."
In the US, around 35,000 people die from the flu every year: worldwide it claims between 250,000 and 500,000 lives a year.
Shaman now wants to test the model with other regions of the US, since:
"There is no guarantee that just because the method works in New York it will work in Miami."
The Department of Homeland Security also provided funds for the study.