Photo: Twitter Health
The University of Rochester’s Adam Sadilek and his colleagues conducted a Twitter experiment.Like Google Flu, they used Twitter data to try and predict when New Yorkers would fall ill.
They were successful.
After examining 4.4 million tweets from more than 630,000 New York Twitter users in 2010, they could predict when someone would get sick up to eight days prior with 90% accuracy.
“Since a large fraction of tweets is geo-tagged, we can plot them on a map, and observe how sick and healthy people interact,” Sadilek writes. “Our model then predicts if and when an individual will fall ill with high accuracy, thereby improving our understanding of the emergence of global epidemics from people’s day-to-day interactions.”
His algorithm can read tweets and decipher when someone is complaining about actual symptoms versus a writing an expression (i.e. “Ugh, that makes me sick”).
The data could make for some cool iPhone apps, but do people really want to know they’re going to get sick? You can’t change something you’ve already caught. With Sadilek’s information, you’d spend eight days dreading what’s to come and stocking up on soup.