Hospitals looking to predict the severity of this year's flu season have a new place to turn: Google.
The Google Flu Trends web site tracks the spread of flu in 20 countries around the world. The site — part of Google.org, the search giant's philanthropic arm — uses search data to see which people in which areas are Googling information about the flu and its symptoms. Since Google logs the IP address of its users, the searches can be matched to physical locations. The more people that search for health information on flu symptoms, the more likely an area is about to experience a major flu outbreak.
The data on Google Flu Trends is so accurate and timely that hospitals should be able to rely on it more than government-published flu case reports, according to a study published Jan. 9 in the journal Clinical Infectious Diseases.
The study, conducted by researchers from Johns Hopkins University School of Medicine, studied Google Flu Trends data for 21 months between January 2009 and October 2010 and found that there was a "strong correlation" between search data and the number of patients seeking treatment for flu-like symptoms in emergency rooms. Since the study counted patients coming into Johns Hopkins' own hospital in Baltimore, it also proved that Flu Trends is specific enough to predict outbreaks in a single city or hospital.
The data from Google Flu Trends was found to be more accurate and timely than the commonly used reports issued by the Centers for Disease Control and Prevention, which can be weeks out of date by the time they reach hospital administrators. Flu Trends data, on the other hand, is available in real-time.
Lead author Richard Rothman, an emergency medicine physician at Johns Hopkins, said in a prepared release that the results of this study show the promise for eventually developing a standard regional or national early influenza warning system for frontline health care workers. An early warning system could help hospital administrators to adjust their staffing levels or make sure that additional resources are available for potential patients.
Other online tools have also been shown to be a good predictor of flu outbreaks. A 2010 study conducted at Southeastern Louisiana University found that Twitter could be used as a surveillance method to monitor the spread of influenza. In the study, assistant professor of computer science Aron Culotta wrote software that analyzed more than 500 million tweets for flu-related keywords and used the resulting data to forecast future influenza rates.
Also on MNN: