Where Is Flu Occurring? "Google" It!

Here is a link to a Los Angeles Times story that highlights work by Jeffrey Shaman and others on a new influenza prediction model using Google search data. These researchers created a new influenza tracking model that pulls data from the Google Flu Trends project and utilizes methodology from weather forecasting. They applied their model retrospectively to flu data from New York City from 2003-2008 and found that they were able to predict the peak of influenza activity over seven weeks before this actually occurred. 

This new study builds on work by Ginsberg and colleagues published in Nature in 2009 (full text available here). These authors created an influenza-like illness (ILI) surveillance model that utilized Google search data. This model correlated closely with ILI surveillance by the CDC, and predicted ILI nationally 1-2 weeks ahead of the CDC surveillance mechanism. 

The work by Shaman and colleagues illustrates that ILI surveillance models based on search engines such as Google can be refined using techniques from other disciplines (in this case weather forecasting). 

Other technologies such as Twitter and Facebook have also piqued interest as potential resources for disease surveillance, as well as for the real-time dissemination of information in the setting of disasters. 

For certain, influenza surveillance models utilizing internet search platforms such as Google are powerful tools, and as these models become more refined and are validated they will provide useful data that will give the medical and public health communities a real-time "heads up" when influenza activity is occurring. It is unclear how other technologies (such as Twitter and Facebook) can best be leveraged for disease surveillance and health communication, but these modalities also represent powerful potential tools.