Preventive Medicine, a medical journal, has published an article suggesting that tweets indicating certain risky actions are very good proxies for existing statistics on HIV outbreaks. When comparing the location data of the tweets with existing statistics researchers from UCLA and Virginia Tech found a high correlation. Over 550 million tweets were evaluated over a 6 month period, 10 000 of which contained words associated with risky behavior. This is a further step in using social media and big data to predict changes affecting the health. A few years ago Google managed to do something similar with flu outbreaks.