Google Flu Trends fails to track winter flu

The use of social media and other online technologies to track the outbreak and spread of flu is a growing trend.  At the forefront of this movement have been Google, who have monitored search queries to detect the spread of flu via their Google Flu Trends service.

Sadly, it seems that the latest US flu outbreak has shown some flaws in how the service operates.  Nature reveal how comparisons between more traditional surveillance data showed that Google Flu Trends was pretty poor at predicting the spread of flu, causing a massive overestimation in predicted flu levels.

The question is whether the problem represents a temporary hitch or a more permanent death knell to the service.

Understanding what went wrong is key to understanding how serious it was for Google.  Firstly, the flu struck much earlier than normal, and was largely over by Christmas.  This timing, combined with the extra strength of the virus this year caused considerable media coverage.

The belief is that this media coverage caused many more people to search for flu related things on Google than was previously the case, and therefore causing a considerable spike in their predictions for those infected.

This isn't the first time that Google have had to adjust their algorithm.  Back in 2009 the H1N1 swine flu virus caused them to tinker with things, and it seems likely that this latest mishap will prompt a similar change.

Critics say that things like Google Flu Trends will never replace more traditional methods of tracking outbreaks.

“It is hard to think today that one can provide disease surveillance without existing systems,” says Alain-Jacques Valleron, an epidemiologist at the Pierre and Marie Curie University in Paris, and founder of France’s Sentinelles monitoring network. “The new systems depend too much on old existing ones to be able to live without them,” he adds.

Traditional flu monitoring depends in part on national networks of physicians who report cases of patients with influenza-like illness (ILI) — a diffuse set of symptoms, including high fever, that is used as a proxy for flu. That estimate is then refined by testing a subset of people with these symptoms to determine how many have flu and not some other infection.

It should be noted however that on the whole, Google Flu Trends has generally performed very well indeed, and has a strong track record of predicting the spread of flu in many countries around the world.