Tweeting Toward Wellness

A Penn State biologist has developed an algorithm that lets researchers evaluate the contents of Twitter postings in order to monitor how diseases spread. Marcel Salathe had students hand-rate whether 70,000 tweets about a new flu vaccine were positive, negative or neutral, then used the results to teach a computer to do the job. The findings showed more outbreaks of flu in geographic areas where Twitterers were posting negative tweets than in regions where tweets were positive.

Scientific research based on social media is the new black, it seems. Researchers love the technique because Twitterers speak for themselves in real time, without influence from interviewers—though, admittedly, the population sampled can be skewed toward the social-media-adept. (Read: Oldsters are screwed.)

While Salathe’s model can be employed to trace the progress of infectious diseases, an NPR report says he hopes to tweak it to it to monitor an even larger challenge in American health care—infectious behaviors such as overeating and smoking, which have also been shown to spread from friend to friend (though not definitively).

“These kinds of data that weren’t around five years ago,” Salathe told NPR, “are going to revolutionize the way we think about health.”