People may rely on social media such as Facebook to showcase the highlights of their lives, like vacations. But new research suggests the language they use in posts might also help predict depression.
Using sophisticated software, researchers were able to scan social media posts and detect depression months before it was apparent on clinical screening tests.
“Social media has made it possible for people to share a little bit of their daily life with researchers,” said study author Andrew Schwartz, an assistant professor of computer science at Stony Brook University in New York.
“Basically, we used the language people wrote on a daily basis and related it to whether they had a diagnosis of depression,” he explained.
Looking at Facebook posts “was slightly more accurate than standard screening questions in finding depression,” Schwartz said.
So what types of language might reveal someone suffering from depression?
Using first-person pronouns was one of the patterns researchers saw. That means people used “I” or “me” frequently in their social media posts.
Schwartz said people eventually diagnosed with depression often talked about their feelings, physical aches and pains, and being alone.
But he cautioned against trying to diagnose your friends or family based on a couple of social media posts.
“A single post isn’t enough to see depression. We were looking at six months of posts before a diagnosis of depression, so I wouldn’t advocate that people try to judge their friends and family,” he said.
Every year, more than 6 percent of Americans experience depression, the study authors noted. But fewer than half receive treatment for the disorder. These high rates of underdiagnosis or undertreatment suggest that current ways of identifying depression could be improved.
The research team was led by Johannes Eichstaedt, a doctoral student at the University of Pennsylvania.
The investigators accessed the Facebook posts of nearly 700 people who had gone to the emergency department of an academic center, including 114 who had been diagnosed with depression. All consented to sharing their Facebook information and their health records.
The researchers reviewed more than half-a-million Facebook posts to build the depression-detecting software algorithm. They determined the most frequently used words and phrases to identify depression-associated language markers.
Using these language markers, the researchers were able to predict future depression as early as three months before it was documented in medical records.