'Do I have the flu?'

"The activity of influenza in Germany continues to rise significantly," Robert Koch Institute (RKI), Germany's disease control and prevention authority, wrote on its website this month. After going around in the United States since December, the flu is now in Germany. Every week, RKI researchers analyze data that is collected from around 700 doctors in Germany. The data analysis is primarily about characterizing the diseases and improving vaccines. The information is then published with a delay of a few days.

Google, on the other hand, is providing the users of its search engine with a daily update on flu outbreaks. It analyzes search terms and presents the data that is collected graphically on a site created for the project.

"For me, the project is absolutely successful," said Martin Memmel from the German Research Center for Artificial Intelligence (DFKI). "This is a classical exploitation of collective intelligence and an epidemic could possibly be predicted a week earlier."

Search terms, historical data help create model

At the beginning of the "flu trend" project, Google compared its own data from searches on the word, "flu," with the diagrams from the centers for disease prevention and control in the US and in Europe. The result showed very strong similarities. Google's employees developed a model, which with the help of this data, displays current figures on the flu in the United States, Canada, Japan and Europe, with the help of their finding. The tool is also available for other countries, including Mexico, Chile, Russia and South Africa, but the models for these countries are experimental estimations, Google said.

The RKI's Susanne Glasmacher said the data were interesting. "But we don't know why people use the search word, 'flu,'" she added. For example, the Flu Trend showed a peak, when many people were talking about bird flu. But at the time there were no cases of bird flue in Germany, she added. A report in the US journal "Nature" also criticized Google for exaggerating the current wave of flu infections in the United States.

Is "Bieber fever" an illness?

Despite such cases, the researchers at RKI, are working with social medial services. Last year, the first trial of an early warning system with the help of data from Twitter came to an end.

The test showed that in some of the cases, epidemics could, in theory, be discovered faster via the microblogging service than with conventional methods. That was, for example, the case with cholera outbreak in Haiti, Glasmacher said. However, the use of words like "fever" in other contexts causes problems. "Bieber fever" which refers to the craze over teen celebrity Justin Bieber would distort the data analysis as his fans often tweet about "Bieber Fever."

Google's service cannot replace traditional tools for monitoring diseases, which the Internet giant also admits. The project, however, can be seen as a starting point from which other applications could be developed. Seeing trends in advance is interesting for every company, said Memmel of the German Research Center for Artificial Intelligence. The center is also working on projects, in which data, especially from mobile devices, would play an important role.

"That way the location will be easier," he said, adding that they are working together with the police.

Their vision for the future includes changing the reaction of the authorities to stampedes and other catastrophes – by making it possible for them to quickly analyze the information from people at the location.

Until there are new ideas and applications, Google's projection of flu outbreaks is mostly good ad for the Internet giant and a plus for its image. And Memmel said he believes that "it also isn't a lot of work once the algorithm is set."