Study: Google searches can provide accurate flu forecasts

Internet search terms really do provide a more accurate real-time forecast of current flu infections, researchers at the University of Warwick recently found. 

Official reports of influenza infection rates are generally produced and released on a one-week delay, while researchers from Google and the U.S. Centers for Disease Control (CDC) have previously said data can show up on Google with almost no delay.

Critics, however, argue that this is too optimistic and cite a 2013 incident when a news scare resulted in higher searches for flu-related terms on Google and thus, a higher estimate of flu cases than was reported by the CDC.

A study, conducted by Tobias Preis and Suzy Moat of the United Kingdom's Warwick Business School and recently published in Royal Society Open Science, says the naysayers' argument doesn't hold true when a simple model is introduced into the flu forecasting. 

“Our results show that dismissing this data is rather like throwing the baby out with the bathwater,” Preis said. “It’s true that simply using the number of searches as an estimate of flu levels can result in misleading figures. However, simple models can be built to watch out for increases in searches that do not correspond to increases in reports of flu and which use this information to improve upcoming estimates.”

In their study, Preis and Moat used an "adaptive nowcasting" model to monitor the link between Google search data and CDC measurements. That information was then taken into account when estimating current flu levels. 

“Predicting the future in the past is of course much easier than truly predicting the future,” Preis said. “To guard against us using information in our simulated ‘nowcasts’ which wouldn’t have been available at the time, we train our model using data from the first 16 weeks. We then test the predictions on the 17th week and retrain our model using data from the second to the 17th week. This model is used to make a prediction for the 18th week, and so on.”

Moat said although the model isn't fail-proof it does improve the forecasting of the flu overall.

“Just like forecasting the weather, however, sometimes these ‘nowcasts’ are wrong," she said. "Our analysis shows that by using data on Google searches for flu-related symptoms, as well as the historic flu data, the error in these ‘nowcasts’ can be reduced by between 14.4 percent and 52.7 percent.”

Organizations in this Story


Want to get notified whenever we write about Aeras ?
Next time we write about Aeras, we'll email you a link to the story. You may edit your settings or unsubscribe at any time.