Can Google predict disease outbreaks?

When you Google your cold symptoms, you're helping point to patterns in the spread of illness.
When you Google your cold symptoms, you're helping point to patterns in the spread of illness.

In a word, yes. As it turns out, if you head for Google at the first sign of illness in your household, you're definitely not alone.

In 2008, a team at Google realized that search terms such as "cold or flu" and "how to treat the flu" could reliably be used as indicators of flu levels in a given region [sources: Mohebbi, Stefansen]. This discovery was used to launch Google Flu Trends, a tool that uses aggregated Google search data to estimate flu activity all around the world.


To identify the search terms that correlate with flu activity, Google worked with the U.S. Centers for Disease Control and Prevention, comparing the popularity of individual Google search queries with historical data about flu outbreaks [source: Google Flu Trends].

Once they identified the most closely correlated search terms, Google researchers determined that spikes in flu-related Web searches actually preceded the flu outbreaks reported by the CDC by about two weeks, suggesting that Web activity was not merely tracking flu activity, but might in fact be able to predict regional outbreaks.

In 2011, Google used data provided by the Singapore Ministry of Health and the World Health Organization to create Google Dengue Trends, a tool that works in the same way to help public health officials predict and prepare for outbreaks of the mosquito-borne dengue virus [source: Sahai].

A study published in 2013 in the journal The Lancet confirmed that Internet-based surveillance can detect outbreaks of infectious diseases such as influenza and dengue fever one to two weeks sooner than traditional reporting methods, which rely on health professionals informing authorities such as the CDC or the WHO only after patients have identified their symptoms and come in to seek treatment [sources: Milinovich, Nauert, QUT].

The Lancet study also revealed that search data was able to detect the SARS outbreak more than two months before any WHO publications, but as of August 2014, Google has no plans to add reporting systems for diseases other than influenza and dengue fever [sources: Google Flu Trends, Nauert, QUT].

Of course, Google and other Internet-based approaches have their limits, the most obvious being that they work only in areas with Internet access and a large population of Web users [sources: Milinovich, QUT]. Web searches are also susceptible to false alerts caused by non-illness-related queries, which could be prompted by factors like medication recalls or news reports about outbreaks in other regions. However, Google is continually updating its model as it finds ways to identify and account for these outside influences [sources: Ginsberg, Stefansen].


Lots More Information

Related Articles

  • Ginsberg, Jeremy, et al. "Detecting influenza epidemics using search engine query data." Google Flu Trends. Feb. 19, 2009. (Aug. 4, 2014)
  • Google Dengue Trends. (Aug. 4, 2014)
  • Google Flu Trends. (Aug. 4, 2014)
  • Google Flu Trends. "Frequently Asked Questions." (Aug. 4, 2014)
  • Google Flu Trends. "How does this work?" (Aug. 4, 2014)
  • Milinovich, Gabriel J., et al. "Internet-based surveillance systems for monitoring emerging infectious diseases." The Lancet. Nov. 28, 2013. (Aug. 4, 2014)
  • Mohebbi, Matt. "Mining trends in search data with Google Correlate." The Official Blog. May 25, 2011. (Aug. 4, 2014)
  • Nauert, Rick. "Data Mining Google May Help Prevent Disease Outbreaks." Psych Central. Jan. 21, 2014. (Aug. 4, 2014)
  • Queensland University of Technology (QUT). "Early Warning: Internet Surveillance predicts disease outbreak." Jan. 17, 2014. (Aug. 4, 2014)
  • Sahai, Vikram. "Using search patterns to track dengue fever." The Official Blog. May 30, 2011. (Aug. 4, 2014)
  • Stefansen, Christian. "Flu Trends updates model to help estimate flu levels in the US." The Official Blog. Oct. 29, 2013. (Aug. 4, 2014)