Researchers, led by Adam Sadilek of the University of Rochester in New York, are looking for a way to use Twitter to predict when individuals will get sick with the flu. By analyzing the 4.4 million tweets with GPS location data from more than 630,000 Twitter users around New York City, the team created a heat map of where people were unwell. They then created a video mapping the spread of illness across the city over the course of a day. Based on that data, the team could predict when an individual would get sick up to eight days before symptoms appeared with 90% accuracy.
Scientists have been looking for more accurate ways to use tech and computers to predict disease outbreaks for years. This new prediction model isn't perfect either. Although it can distinguish between tweets of "I feel so sick" and "I'm sick of this traffic," it cannot account for people who do not reliably tweet about their flu symptoms. Also, the system only measures location-based contact with other sick people, which is not the only way to pick up a bug.
More from Tecca: