Australasian Science: Australia's authority on science since 1938

Disease Outbreaks Identified with Google

By Stephen Luntz

Infectious disease outbreaks can be detected 2 weeks earlier by tracking internet searches, a review in Lancet Infectious Diseases has confirmed.

Many people first respond to symptoms by going online, particularly for symptoms they have not experienced before. Google.org, the search company’s philanthropic arm, compared eight million search terms with outbreaks of influenza and found 165 that could be included in a model. “They had to check those that correlated,” says lead author Dr Gabriel Milinovich of the University of Queensland’s School of Population Health. “They initially found a correlation with ‘college basketball’ because most outbreaks in the US occur in the basketball season.”

A number of studies have investigated whether this model can be used to predict successful outbreaks, and whether the same approach works for other diseases. “We initially tried to look at all diseases where there have been studies, but we realised most were on ‘flu and a few on dengue so we focused on those,” Milinovich says. “Both are significant globally and for Australia.”

Traditional reporting methods take time as patients may not go to the doctor immediately and even then may take a while to get an appointment. While legislation requires doctors to report suspicions of some diseases, in other cases they will send swabs or blood samples off to labs and wait for the results, creating even longer delays. Internet searches circumvent this.

Nevertheless, Milinovich says that the approach is not without problems. Google.org has refused to release the search terms that go into its model, which Milinovich thinks is “for IP reasons” rather than fears that trolls will game the model to waste resources. “Governments shouldn’t be dependent on corporations, even their philanthropic arms,” says Milinovich. “If this is to be used in future, public health authorities need their own model.”

To that end his team is working to find the terms that can be used to predict other diseases. They are also investigating whether social media could provide an additional avenue to alert authorities, since every day gained can prove lifesaving in the face of an outbreak. A news aggregator picked up evidence of the disease that became known as SARS 2 months before the World Health Organization.

“A net-based system will not replace traditional methods,” says Milinovich, “but allows an opportunity to send a public health unit in to investigate if the problem is real”.