New Study Predicts Depression Based on Types of Images You Post to Instagram
If social media isn’t guilty of causing depression, it could now be the thing that helps diagnose it. At least, that's according to a new study that claims to have found a way to diagnose mental health issues by looking at the sorts of pictures users post to their Instagram accounts.
The research, conducted by a team from Harvard and the University of Vermont, used a set of machine learning tools to build a model that could detect early signs of clinical depression with up to 70 percent accuracy. How, you may skeptically wonder?
By taking 166 different users, who together had posted 43,950 photos, the model made its predictions based on a data review that took into account "color analysis, metadata components, and algorithmic face detection” of each user’s feed. In simpler terms, the team was able to build a model that could identify a set of variables—such as types of filter used, or facial expressions—that were consistent with signs of depression. As it turns out, their model had a greater diagnostic success rate than a general practitioner.
This is, of course, far from a fool-proof system. For instance, in order for this model to work, a user needs to be posting images to Instagram semi-regularly. But, as the paper noted, as social media becomes increasingly ingrained in people’s lives, their model could eventually "serve as a blueprint for effective mental health screening in an increasingly digitalized society." Just imagine the implications this could have for Google ad algorithms.
As stereotypical as this may sound, the study found that depressed users generally opted for blue-gray or black-and-white filters on their images, with happier grammers going for warmer colors. The paper had nothing to say about Instagrammers who used #nofilter.