The flu hit especially hard this year in the northern hemisphere, but Google’s Flu Trends map might not be a good indicator of just how bad it was. As Nature points out, a comparison between Google Flu Trends (which bases its map off flu-related searches) and traditional surveillance data showed that our beloved search engine had grossly overestimated the extent of the flu epidemic.
Google’s algorithms will of course improve with time, but this little fluke acts as a reminder that data gleaned from the internet and social media won’t really ever be a reliable substitute.
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.
And according to Alain-Jacques Valleron, an epidemiologist at the Pierre and Marie Curie University in Paris and founder of France’s Sentinelles monitoring network:
It is hard to think today that one can provide disease surveillance without existing systems. The new systems depend too much on old existing ones to be able to live without them
Google’s computerised surveillance is generally pretty accurate; Nature notes that researchers in a number of countries were able to confirm that its influenza-like illness estimate matched up to traditional survey results. However, as they depend on people’s searches, the nature of Google’s methods makes them victim to the unpredictability of human nature.
Google would not comment on this year’s difficulties. But several researchers suggest that the problems may be due to widespread media coverage of this year’s severe US flu season, including the declaration of a public-health emergency by New York state last month. The press reports may have triggered many flu-related searches by people who were not ill. Few doubt that Google Flu will bounce back after its models are refined, however.