EPJ Data Science Highlight - Are your tweets feeling well? Opinion and emotion in tweets change when you get sick
- Published on 02 July 2017
Can we tell if a person is physically ill by the way they tweet? On a recently published article in the journal EPJ Data Science, researchers at the Pacific Northwest National Laboratory uncover links between the health of users and the emotional tone of their social media output.
Guest post by by Svitlana Volkova, originally published on SpringerOpen blog
Any doctor or nurse knows good public health begins with prevention. Whether it’s a severe strain of the flu or mental illness, identifying the need for help early can save lives. Social media could be the game-changing solution public health workers have been looking for. Whereas traditional data from clinics may take weeks to collect, social media streams in real time. In other words, public health workers could monitor social media like a heartbeat, and take action before people visit a doctor.
Public health trends on social media are more nuanced than looking for spikes of “I feel sick” or “flu.” To truly tap this source of public data, we need to understand patterns of how people behave differently on social media when they are sick. We believe that expressions of opinion and emotion may be that signal.
The Department of Energy’s Pacific Northwest National Laboratory studied 171 million tweets from users associated with the U.S. military to determine if the opinions and emotions they express reflect visits to the doctor for influenza-like illnesses. We compared military and non-military associated users from 25 U.S. and 6 international locations to see if this pattern varies based on location or military affiliation.
Overall, we found that behavior significantly varies by location and group. For example, tweets from military populations tend to contain more negative and less positive opinions, as well as increased emotions of sadness, fear, disgust and anger.
The baseline is fuzzy, and that should be no surprise. People behave differently based on the world around them. To that end, we identified location-dependent patterns of opinion and emotion that correlate with medical visits for influenza-like illnesses. And a general trend did appear: Neutral opinions and sadness were expressed most during high influenza-like illness periods. During low illness periods, positive opinion, anger and surprise were expressed more.
Opinion and emotion may not be the strongest predictors of illness, but they offer a unique measurement. Many studies using social media rely on health-related text, where health is measured based on the presence of specific words. Opinion and emotion are present in every tweet, regardless of whether the user is talking about their health. The signal is more subtle and nuanced, but we’re discovering opinion and emotion act like a constant digital heartbeat.