Google is tracking influenza outbreaks, and it’s actually pretty cool

In this interconnected world, companies are finding out that you can reliably track the spread of infectious diseases by what search terms people are using. One really cool example of this is Google Flu Trends whereby Google uses aggregated Google search data to estimate current influenza activity around the world in near real-time. Google’s Flu Trends is a product of the “big data” excitement in the mid 2000s. Since 2008, Flu Trends tracks 45 flu-related search terms (Google won’t specify which ones) over billions of searches in 29 countries. While the CDC also does influenza surveillance and provides data, that data has a delay of a couple weeks so it’s not totally reflective of influenza in real-time.

An article in Science back in March 2014 reported that Google’s Flu Trends overestimated the number of flu cases over the last couple years. Over the past couple years the Google numbers have consistently been higher than the CDC reported numbers. While Google has updated its algorithm, they still overshot by 30% for the 2013-2014 flu season. However, the important part of the article, which didn’t get as much mention as how off the flu predictions were, was that this tool was never designed to supplant the surveillance the CDC is doing. The Lazer paper showed that when you combine Google’s Flu Trends data with the CDC data, you get a better result that you would just using either one. The folks at Google who designed Flu Trends got in touch and worked with the CDC right from the start in an attempt to build a complementary signal to other signals, meaning that the creators didn’t want to combine their data with the CDC’s as then it couldn’t be a separate way of understanding the influenza data.

Google has expanded this “big data” concept to include tracking Dengue activity around the world. Dengue Trends encompasses 10 countries and uses the same idea as Flu Trends, tracking Dengue related search terms. The reason this expansion into including Dengue, when it isn’t currently a disease endemic in the US, is important is because Dengue is a growing threat to public health in the tropics and is thought to kill at least 20,000 people a year. Dengue is endemic in India but the official case reports are hopelessly inadequate. Based on newspaper reports and word-of-mouth, it’s clear that Dengue is an incredibly common affliction but official reports only state 12,500 cases in a country of 1.2 billion. A new report (October 2014) estimates that the true number of Dengue cases in India are 282 times greater than official number. Google’s Dengue Trends provides another method for tracking Dengue that may offer a more realistic representation of cases where the current surveillance capacity and methods may not be accurately reflecting information that public health officials need for intervention and prevention efforts.

Leave a Reply

Your email address will not be published. Required fields are marked *