Twitter as a tool in epidemiology

If you think no one cares when you tweet about being sick, think again. Researchers at Johns Hopkins University have found a way to filter those annoying messages to turn them into something useful. While researchers have been using data from Twitter and other social media outlets for a while now, it was often difficult to sort through the noise. After all, Twitter facilitates more than 340 million tweets per day.

The researchers have developed algorithms to help sort tweets about the flu. After all, some people may just be talking about the flu because it's in the news. Perhaps they're sharing an article. The researchers found that relevant tweets containing the word 'flu' are often accompanied by other keywords like fever, Tamiflu and doctor. This filtering helped sort tweets about influenza from other 'maladies' picked up by Twitter like Bieber fever (though one can argue that is equally disturbing and certainly needs to be contained).

This real-time information can help hospitals prepare with more beds or let schools know when to close as a preventative measure. Researchers aren't satisfied with just seeing where the outbreaks are; they want to be able to predict when and where illnesses will spread.

The Centers for Disease Control (CDC) provide valuable information in their weekly reports including the number of cases that required hospitalization, the number of deaths and a breakdown of strains of the flu virus. This information is much more scientific and lab based than pulling information from Twitter. Unfortunately, by the time the CDC releases a report, the statistics are about two weeks old.

The collection of Twitter data is lacking in some ways. Most notably, young children and the elderly are not likely tweeting their symptoms.  Even so, during this season's flu outbreak, their statistics matched up fairly well with those from the CDC. Researchers hope this new technology can be used to track other diseases as well.

Do you post about your illnesses on Twitter or Facebook? Tell me how you feel about researchers using your updates in this manner.

Twitter logo sketch copyright of cbhdesign via flickr.