Unprecedented new research uses billions of tweets to reveal surprising patterns about cancer, obesity, and other ailments
Increasingly these days, when people get sick, they're announcing it to the world using Twitter. It might seem self-indulgent, but it's also a practice that may revolutionize public health research.
In an unprecedented study, Johns Hopkins computer scientists Mark Dredze and Michael J. Paul have analyzed 1.6 million health-related tweets to uncover health trends that are consistent with official government data, plus new findings about self-medication, localized disease outbreaks, and more. They collected 2 billion public tweets posted between May 2009 and October 2010 and used a new proprietary algorithm called ATAM+ to sift through the ones related to health by using keywords linked to various ailments, symptoms, and treatments.
(Story continues below)
"Looking at an individual tweet might not say much. It's a mess," Paul says. "But if millions of people are saying the same thing, they do tell a story that might be meaningful."
Dredze and Paul's study is the first to use large quantities of social-media data to find trends related to a broad range of ailments. Previous works focused only on specific ailments, such as the flu.
The researchers are slated to present their work next week in Barcelona at the International Conference on Weblogs and Social Media. Still, they're hoping their research will ultimately appeal to a whole other audience: healthcare professionals.
"It's not a magic tool. It's a dialogue," Dredze says. "Now we want them to ask specific questions they'd like answers on so we can find health information that's useful for everyone."
In the gallery above, Dredze and Paul share with The Atlantic 10 intriguing patterns they found about everything from influenza to insomnia that illustrate how social media could play a big role in the future of healthcare research.
Image: Michael J. Paul
We want to hear what you think about this article. Submit a letter to the editor or write to email@example.com.