Social Networks For Disease Tracking

Automated disease tracking, sometimes known as Biosurveillance, is the process of collecting and analyzing a variety of forms of data to determine disease trends. I briefly worked in this field, which makes me believe I know something about it. For a while now, I've been considering the utility of status updates posted on social networks as a potential source of public health information, and it looks like I'm not the only one.

While I hadn't gotten around to actually writing some code to scrub Facebook or Twitter, others have been much more productive. Lampos, et al from the University of Bristol have been using tweets to determine epidemiologic trends with a decent correlation to data collected from the UK's HPA (Health Protection Agency).

The power of social network data for biosurveillance comes from the fact that it is readily available in real-time. Traditional sources of epidemiological data include over-the-counter drug sales, public school absentee counts, and emergency room chief complaints. Most of these sources take days or weeks to compile and report. Chief complaints can be gathered within hospitals quickly, but most are not reported to disease tracking agencies for weeks or even months. Some companies make a living contracting with hospitals to collect and analyze such data as quickly as possible to report near real-time disease trends, but even this approach is limited in that it requires legal agreements and a data reporting infrastructure to be set up between the data collection center and each hospital to be included in surveillance. This is no small task, considering the fact that the U.S. alone contains 5815 registered hospitals as of 2008 and the inclusion of community hospitals puts the number well over 10,000. In addition, emergency room surveillance represents only the fraction of the population moved to visit a hospital, likely missing the earliest data indicating an emerging epidemic ("I feel sick", "I can't go to work", "Let's skip the Murphy's cookout, they always serve vegi-burgers anyway").

The modern world of social networks drives people to constantly report their status to the web, with little concern for who, exactly, is listening. The upshot is that we can use this compulsion for chatter to gather much more data than ever before and determine health trends faster than we could have dreamed of a decade ago. Combine this with other emerging technologies such as Google's efforts to apply search data to public health trending, and we have some seriously powerful tools to predict and track epidemics in their earliest possible stages. This is great news for those who like to be healthy, as the response time and planning capabilities of hospital systems and public health agencies stands to be dramatically improved in the future as this technology takes hold.