If you hang around or come into contact with a lot of sick people, you’re very likely to fall ill yourself. Common sense, perhaps, but what if we could use social media to predict the likelihood of that happening in advance?
Researchers at the University Of Rochester in New York have used Twitter to track the outbreak of flu through New York, and, using learning model, have been able to determine when healthy people are about to fall ill with an accuracy level of some 90 percent.
The study, undertaken by Adam Sadilek and his team, analysed 4.4 million tweets that contained GPS location data from some 630,000 users in New York City over one month in 2010, using an algorithm that learned the difference between actual reports of illness and other, non-relative uses of words such as “sick”.
The results were plotted on a heatmap, which can be used to predict when people in a certain area were at risk of contagion up to eight days in advance.
See the video below for a visualisation – the more red an area is, the more people are affected by flu in that location.
“Given that three of your friends have flu-like symptoms, and that you have recently met eight people, possibly strangers, who complained about having runny noses and headaches, what is the probability that you will soon become ill as well? Our models enable you to see the spread of infectious diseases, such as flu, throughout a real-life population observed through online social media,” says Sadilek.