As you click back and forth between the Juno and Clarendon, wondering which Instagram filter will make your lunchtime salad look most appealing, you probably aren't considering what a filter might say about your mental health. As it turns out, your filter choice, along with a trove of other Instagram data, may hold the key to reliably diagnosing depression.
In a recent study, researchers Andrew Reece of Harvard University and Chris Danforth of University of Vermont built an algorithm which looked at patients' Instagram feeds. They were able to diagnose depression with an accuracy rate of 70%. Their algorithm is better than a general practitioner's assessment of depression after a face-to-face visit, which is only about 50% accurate.
This isn't the first time an algorithm has used social media to diagnose a user's mental health. But this study's algorithm is new and groundbreaking in that it uses images to do so. Tweets and Facebook status updates can be analyzed for content, frequency, and engagement, but Instagram posts have an additional aspect highly correlated to mood: color.
What they found will likely come as no surprise: photos posted by depressed individuals were more likely to be blue, gray and dark, while those with more vivid colors indicated happier users. Beyond that, they saw that depressed participants were less likely to use any filter on their photos at all. When they did use them, there were clear trends. Inkwell overuse, for example, may not just be a sign of pretension, but an actual cry for help.
Something as seemingly trivial as filter choice may be a good indicator of mental health precisely because of its banality. Filters are quite literally overlaying how you want see the world on top of how your camera has captured it, without the intrusion of anything so messy or well considered as words. Instagram users also tag their photos with filter hashtags, which further back up these findings with some empirical data. There are far more photos of smiling faces under the Valencia hashtag than there are under Inkwell hashtag.
Faces, in fact, are another predictive attribute of Instagram posts. The algorithm used facial recognition to chart how many people were in photos, finding that depressed participants were more likely to post photos with faces, but that their photos had a lower average face count than healthy participants.
Finally, the researchers also looked at engagement; both the users engagement with the platform, and their followers engagement with each post. Users with a higher posting frequency were associated with depression, though this wasn't as reliable a predictor as the content cues of color or number of faces. They also found that the more comments a post received, the more likely it was posted by a depressed person. The opposite is true when it comes to likes, unsurprisingly: fewer likes meant sad posters.
While knowing that a machine could determine your mental health based entirely on your social media profile may be terrifying, there could be some real benefits from this technology. In addition to its remarkable accuracy, there's the associated time and cost savings of avoiding wrong diagnoses. The other interesting advantage is access. While many people may have a hard time getting to the doctor for financial reasons, almost everybody has an Instagram feed. Now we just have to decide who gets access to them.
Cara Rose DeFabio is a pop addicted, emoji fluent, transmedia artist, focusing on live events as an experience designer for Real Future.