Google watches over fluBMJ 2008; 337 doi: https://doi.org/10.1136/bmj.a3076 (Published 31 December 2008) Cite this as: BMJ 2008;337:a3076
- Geoff Watts, freelance journalist
With every search of the internet we reveal ourselves: where we are, what we want to know, what we need. It all adds up to billions of bits of information that can help corporations locate us, tempt us, and sell to us. As the interests served by collecting and analysing these revelations are not ours, it’s heartening to learn that the owner of one of the biggest of the information banks, Google, has found a benign use for some of the search data squirrelled away in its vaults. By tracking our use of certain computer search terms it can reveal trends in the incidence and spread of influenza1—and do so at a speed that leaves conventional reporting methods way behind. But is this just a curiosity? Or could this unconventional method of disease surveillance hold real public health benefits?
The lead engineer on the Google flu trends project is Jeremy Ginsberg, who works with a team that studies how people search for information. Their assumption was that a proportion of those who develop a flu-like illness would go to the web for advice on what to do. But whether an attempt to use aggregated search data in modelling the emergence and spread of flu would succeed they had no idea.
Their first task was to choose the search terms that people would be most likely to use.2 Anyone already believing that that their aches, pains, and snuffles are due to the flu might well type in just that word. But others might instead enter, for example, one of these symptoms—each of which can …