Google uses search patterns to accurately estimate flu activity

Computerworld - Google Inc. has found that analyzing aggregated Google search data can estimate future flu activity in a state as much as two weeks faster than traditional government disease surveillance systems.

The company published its findings this week as Google Flu Trends, which noted a close relationship between how many people search for flu-related topics and how many people actually have flu symptoms.

Observers said the ability to predict disease outbreaks could boost Google's health care efforts. The company launched its Google Health online personal health record to consumers in May after announcing, to great fanfare seven months earlier, its intent to enter the health care market.

In its flu trends report, Google compared its queries with data from a flu surveillance system managed by the federal Centers for Disease Control (CDC) and found that certain search queries become more popular during the flu season, the company noted in a blog post. By counting these queries, Google said it can estimate how much the flu is spreading in various regions of the country.

During the flu season last year, Google compared its search results with the CDC's data and found that it accurately estimated flu levels one to two weeks earlier than the federal agency.

"It turns out that traditional flu surveillance systems take one to two weeks to collect and release surveillance data, but Google search queries can be automatically counted very quickly," Google said in the blog post. "By making our flu estimates available each day, Google Flu Trends may provide an early-warning system for outbreaks of influenza."

For epidemiologists, Google added, this means that using Google to detect flu outbreaks earlier could reduce the number of people affected.

"If a new strain of influenza virus emerges under certain conditions, a pandemic could emerge and cause millions of deaths," Google added. "Our up-to-date influenza estimates may enable public health officials and health professionals to better respond to seasonal epidemics and -- though we hope never to find out -- pandemics."

Google did note that Google Flu Trends cannot identify individual users because the company uses anonymous and aggregated counts of how often certain search queries occur over a week.

Rick Turoczy, a blogger at Read Write Web, noted that while the flu is the first target of this experiment, it potentially could be used to predict outbreaks of other diseases.

"More importantly for Google, coupling this kind of anonymous aggregated data with other Google offerings could further the company's moves into the health care space," he added. "Just imagine, in the not-too-distant future, you could be warned of potential disease outbreaks in your city when logging into your personal health record on Google Health. It's not a huge intuitive leap, but it's a leap that puts the responsibility for health in the hands of the individual.

Read more about internet applications in Computerworld's Internet Applications Knowledge Center.