If you are experiencing depression, please call Lifeline on 13 11 14 or Beyond Blue on 1300 22 4636. Help is available 24/7.
The connection between social media and mental health isn’t generally seen as a positive one. Just this May a survey found Instagram to be contributing to anxiety, depression, bullying and fear of missing out in young users.
But research out of the US suggests that the image-sharing platform could, in fact, be a useful tool for identifying if someone is struggling with their mental health.
In a study published in the journal EPJ Data Science, researchers from Harvard University and the University of Vermont suggested that it’s possible to identify a user with a history of depression by analysing their Instagram feed.
The team used machine-learning tools and image processing to examine 44,000 photographs posted by 166 study participants. These 166 people were either clinically ‘healthy’ or had received a diagnosis of depression in the past.
In doing so, researchers found that a number of revealing traits tended to be shared by the latter.
For one, depressed individuals published images at a higher frequency, but while those posts received a higher number of comments, they received fewer likes. Depressed users also tended to post fewer faces per photo than their 'healthy' counterparts.
Colours and filter use were also telling.
“People in our sample who were depressed tended to post photos that, on a pixel-by-pixel basis, were bluer, darker and greyer on average than healthy people,” study co-author Andrew Reece, a postdoctoral researcher at Harvard University, told The New York Times.
So while the healthy participants tended to used filters like Valencia that lighten and brighten their images, the others were more likely to favour Inkwell, which turns them black and white.
LISTEN: Things we've all done on Instagram. (Post continues below.)
While the authors acknowledged that their study was small - 166 participants isn't exactly a significant sample size - they believe the results speak to the promise of the technique; that a machine-learning model could one day prove useful in conducting or contributing to mental health screenings.
"Given that mental health services are unavailable or underfunded in many countries, this computational approach, requiring only patients’ digital consent to share their social media histories, may open avenues to care which are currently difficult or impossible to provide," they wrote.
As co-author Christopher Danforth, a professor at the University of Vermont, told The New York Times, “We reveal a great deal about our behaviour with our activities, and we’re a lot more predictable than we’d like to think.”