‘Slice of life’ tweets have been some of the most scorned content on the Internet. Who really cares if you’re frying up grass-fed bacon by the pound or binge watching the latest season of House of Cards from your couch? Most of us consider this the custard-like filling of the Twittersphere—lots of calories, little substance.
However, by virtue of sheer volume, these very tweets may be useful for tracking and forecasting health-related behavior if the data can be extracted in an accurate and efficient manner. Increasingly, ‘big data’ innovators are harvesting the 200 billion tweets posted each year to help inform and influence public health efforts in a growing field known as computational health science.
Take the Lexicocalorimeter for example. Researchers at the University of Vermont developed this online, interactive tool to measure the caloric intake and output of Twitter posts by building an extensive list of foods and activities and assigning each a number of calories. The rough ratios of these measures are presented by state to establish a real-time ranking of caloric balance.
Generated by the Lexicocalorimeter, the maps below show which food and activity was most significant for each state at a given point in time. For example, “tomatoes” and “dancing” lead in California while “cake” and “eating” are most popular in Mississippi, the most obese state in the nation. Turns out the tool’s caloric balance data strongly correlates with health stats reported by the CDC’s Behavioral Risk Factor Surveillance System—the gold standard of behavioral surveillance.
There also have been a number of efforts to use social media to track and predict the magnitude and progress of the flu. During the 2012-2013 flu season, scientists from Johns Hopkins University and George Washington University developed a Twitter tracking system that was 93 percent accurate when compared to national flu data collected by CDC (“National and Local Influenza Surveillance through Twitter: An Analysis of the 2012-2013 Influenza Epidemic”).
To achieve this level of accuracy, researchers had to create an algorithm that would separate chatter from useful information. For example, one problem with mining Twitter to determine flu incidence is that people aren’t just using the platform to discuss their own exposure or symptoms, but also to discuss the flu in general (especially after relevant news coverage). There are also thousands of tweets that need to be weeded out even though they include relevant terms (e.g., “Bieber fever” or “the cost of gas makes me sick”).
While there are opportunities and challenges to consider, both of these examples indicate that Twitter has the power to track and predict public health issues.
We must consider how we can use this information as health communications professionals. For the most part, my day-to-day interaction with Twitter revolves around the content that clients can push out and less on how they can listen and learn from what others are posting. But clearly there is a lot to learn and act upon if we spend more time harnessing the power of Twitter.
To start, we should use this type of data to inform awareness, education, and behavior change efforts by better understanding when and why people are collectively talking about particular public health topics or activities. With insights gleaned from tools like the Lexicocalorimeter, we can design education and outreach efforts with tailored, state-specific health messages; with flu data, we can predict where and when illnesses will spread, providing public health systems with advanced warnings and more time to pull together necessary resources.
Twitter data can also be used to identify misperceptions around health issues, therefore, informing what audiences to target with communications efforts. The Hopkins analysis of flu-related tweets found that a significant number of people were taking antibiotics to treat flu symptoms; however, we know that antibiotics don’t treat the flu, which is a virus. This valuable insight should be used to help inform messaging for flu experts interacting with the media and future antibiotic misuse campaigns.
As communications professionals, we must be nimble, efficient, and constantly innovating to create and refine our outreach strategies. I look forward to following this growing trend as we continue to realize the power of Twitter’s collective voice—even all the content ‘junk food’ that inspires more than the occasional eye roll on my part. Maybe I’ll have more tolerance for it now!