Twitter app takes on disease tracking

The MappyHealth creators have processed some 70 million tweets for disease tracking purposes.

Twitter, the much beloved social networking site, is set to take on disease outbreaks, after the collaborative efforts of three informaticists yielded a new Web-based application tool available to public health officials.

Officials at HHS' Office of the Assistant Secretary for Preparedness and Response (ASPR) announced in September the arrival of MappyHealth, the winning submission in a developers' challenge sponsored by the ASPR.

The triumvirate was awarded $21,000 for developing the application in response to a request made by local health officials for help developing a Web-based tool that could make social media monitoring more accessible to local health departments. 

According to public health officials, studies of the 2009 H1N1 pandemic and the Haiti cholera outbreak demonstrated that social media trends could indicate disease outbreaks earlier than conventional surveillance methods. However, many Web-based apps are retrospective, looking back on a disease outbreak, as opposed to attempting to identify health trends as they emerge in real time.

Brian Norris, co-creator of MappyHealth, says what started out as a fun project turned into a mission to utilize data analytics to improve health. 

"I think what we found as we've gone through this is there are a lot of different ways you can see information." For Norris, the challenge really transformed into building a platform that could be leveraged in healthcare to solve other analytical problems via social media platforms.

When the data analytics portion of the project began, Norris recalls what the group initially observed. 

As the World Health Organization issued health reports, for example on Coronavirus or West Nile Virus, "We were able to see how those reports sort of moved around the social media Twittersphere, and that's valuable information, because it also gives information as to how far out their reports have gone, how long people are actually talking about it." 

Mark Silverberg, another co-creator of MappyHealth, explains how the data analytics portion works. First, the group requests the tweets they're interested in. "Prior to the contest, they gave us a list of key words, but we've filtered it down and added our own." Some of these key terms include a variety of illnesses, such as influenza and malaria. 

Once they're connected to the servers, the health tweets start streaming in, often millions within an hour, Silverberg says. "Right when we receive [the tweets], we do some quick analysis, look to see what condition they're talking about. We apply those qualifier terms," which Silverberg says are things like "I have," or "death," or "I'm going to go see a doctor."