Scientists at Stony Brook University think it might be possible to use social media to look for signs of depression. They’ve built an algorithm that scans Facebook posts to find what they call “linguistic red flags.”
Linguistic red flags – meaning key words you might include in your Facebook post that could let on a lot about your emotions and your mental state.
“Topics related to crying, for example. That was associated with a depression diagnosis,” Stony Brook University’s Andrew Schwartz said.
Schwartz and his team found subjects who’d been diagnosed with depression, and then they asked to read their status updates.
“We saw people also talking about missing people or being lonely. Even talking about somatic complaints, like my head hurts, I have a headache.”
They used that data, things like those keywords from before, to build an algorithm they said can predict with surprising accuracy whether someone might be suffering from depression.
“You can get a prediction that we found on average was slightly better than the screening questionnaires used in clinical settings.”
Schwartz said the idea that you can study someone’s word choice to get a sense of their mental health is a fairly old one. Some of the first mental health questionnaires did this by asking people to free-associate words or write journal entries.
“With the rise of social media, it’s now become possible not just to ask people in one sitting to write, or to ask people to fill out a questionnaire about words, but rather to literally look at the words they’re using.”
Of course, an algorithm is no substitute for a doctor’s diagnosis when it comes to mental health. But Schwartz says doctors could use this algorithm as a tool to help them make their own diagnoses.
The work was done in collaboration with Johannes Eichstaedt and Raina Merchant at the University of Pennsylvania.