There are certain telltale signs that seem to pop up whenever people are about to get sick, like feeling achy, tired, and generally a bit off. But predicting when you’ll get sick before symptoms start? The human body isn’t that sophisticated. Twitter might be.
Adam Sadilek and his computer science colleagues at the University of Rochester have come up with a way to predict when people will get sick eight days before they fall ill, with a little help from Twitter, machine learning, and natural language understanding techniques.
As New Scientist explains, Sadilek’s team taught a machine-learning algorithm to sift through 4.4 million tweets (all tagged with GPS data) from more than 630,000 Twitter users in New York City over the course of a month. The algorithm could distinguish between people who were talking about actually being ill and those who used sickness-related words in other contexts–like saying they were “sick” of a certain song, for example. The result: The algorithm could figure out when healthy people would get sick up to eight days ahead of time, with an accuracy rate of 90%.
Check out a heat-map visualization of sickness in New York City in Sadilek’s video above. As his website explains: “We show emergent aggregate patterns in real-time, with second-by-second resolution. By contrast, previous state-of-the-art methods (including Google Flu Trends and government data) entail time lags from days to years.”
Sadilek and his team aren’t the first to try to use social media to track illness in real-time. Sickweather, a startup that gathers data from Facebook, Twitter, and other social networks to figure out what illnesses are spreading around your area, also tracks sickness-related keywords to monitor where the flu or the cold is being passed around.