Twitter Feed and Spread of Disease

Did you know that researches have found a correlation with the real time Twitter feed and the spread of disease?  Apparently its not a brand new concept either.  However, before researchers were focusing more on trending topics and they found nothing but white noise.  The Twitter feed created a rampant paranoia that spread well before the spread of disease effected the masses.  After the spread of disease, the Twitter feed actually slowed down and there was less hype surrounding it, at least in the case of H1N1.  But wait, researchers had it all wrong when deciding to focus on trending topics, see how they revolutionized their method to make the Twitter feed a worthy research tool when tracking the spread of disease.

Twitter Feed and the Spread of Disease | Going Beyond The Trend

Twitter isn’t the first online resource used to track the spread of disease, as a matter of fact people at Google and Yahoo found indicators in search terms that helped with research, thus birthing the idea for researches to analyze the Twitter feed later on:

Researchers from Google and Yahoo had already found that certain search terms were good indicators of flu activity. Google had even launched Google Flu Trends, which provides public estimations for flu activity. But Twitter, it seemed, was only distracting us from reality — not helping us understand it.

Yet a multi-disciplinary team of researchers at the University of Iowa had hope that Twitter could not only track the reaction to H1N1, but also track the disease itself by using contextual information in tweets that isn’t available in search terms.

“We had no idea that we would actually be able to do the second part when we started it,” explains Dr. Philip Polgreen, one member of that team. And so, it seems, Twitter could become an innovative scientific tool for epidemiologists.

You see what these researches did was locate terms and changes in the Twitter feed as related to the fluctuation in the number of cases reported to the Centers for Disease Control.  This is a totally different approach than the method used by Google and Yahoo, in that they weren’t searching for popular search terms or trends, but rather a change in dialogue across the Twitter feed to track the spread of disease.

The program located Twitter terms that fluctuated in relation to actual cases, which were often related to aspects of disease such as fever temperature and doctor visits.

It then used those terms to spot signs of the flu in other geographic areas before cases were reported. The accuracy of the results demonstrated that it was possible to cut through Twitter’s noise and locate indicators for the flu. At the time, the real-time estimations in the study were one to two weeks faster than that of traditional flu tracking mechanisms — a time difference that matters greatly in making disease estimates useful.

Twitter Feed and the Spread of Disease | A Long Way To Go

Although these techniques are very interesting and hopeful, tracking the spread of disease through the Twitter feed is still a work in process and there are kinks to be fixed.  For example not everyone uses Twitter, therefore in some regions this wouldn’t be effective.

“I don’t think this will replace traditional surveillance at all,” Polgreen says. “In fact, there’s no way to validate these lines of investigation without traditional surveillance information. But I think, and I like to hope, that it provides another stream of information.”

That being said, the Twitter feed is nonetheless a great asset in the process of tracking the spread of disease.

Image Attribution: http://www2c.cdc.gov/podcasts/player.asp?f=11226