Algorithm detects depression levels among Instagram users

Machine-learning techniques help paint mental health portrait from 40,000 images

Instagram posts: while photos from depressed users featured more people than the average user, there were fewer visible faces present.

Instagram posts: while photos from depressed users featured more people than the average user, there were fewer visible faces present.

 

An algorithm developed by researchers at Harvard and the University of Virginia in the US can detect Instagram users suffering from depression. There is a reason they call it the blues: users with depression are more likely to post images in shades of blue, grey and black. The study found that depressed users favoured the black and white “Inkwell” filter while healthy participants preferred ‘”Valencia”, a filter that brightens images.

Depression diagnosis

It was found that while photos from depressed users featured more people than the average user, there were fewer visible faces present. Their photos also typically had fewer “likes” and fewer comments. This kind of machine-learning research may have future applications as a diagnostic tool; the authors say it could “serve as a blueprint for effective mental health screening in an increasingly digitalised society”. arxiv.org/abs/1608.03282