Social media has been on the radar for the public health workers by means of its potential to offer real-time access to millions of short, geographically localized messages. Since those messages also include information about the health status of the social media users, it is not a surprise that we could manage to make a faster and more efficient surveillance with those messages.
One of the obstacles to make such a surveillance system work sufficiently is that the data algorithm that we develop should be able to differentiate between the chatter and the infection that has been mentioned in the messages. An algorithm that has been developed by Broniatowski et al. has managed to develop such an algorithm on Twitter. They have filtered all the relevant tweets from October 2012 to May 2013 and they have seen that their predictions just by using their own algorithm have been consistent with the epidemics report by CDC in the same time interval