We already know that using social media can make us feel crazy. Scrolling through your Facebook feed can trigger feelings of jealousy, loneliness and discontentment — and that’s just on a good day.
It’s also no secret that millennials love using Instagram. It’s the place where FOMO and filters collide to create a carefully curated reality; it’s a fairy tale, but with better lighting.
There’s no doubt that for some people, looking at Instagram can lead to depressive feelings. But what if someone could tell you were struggling, just by looking at your Instagram grid? What if the information we regularly mine from social media could be used to detect mental illness and maybe even save someone’s life?
According to researchers Andrew Reece and Chris Danforth from Harvard and the University of Vermont respectively, there is a strong link between a person’s mental health and the photos they post on social media. To test their theory, the researchers analyzed the mental health of 170 workers from Amazon’s Mechanical Turk service, 70 of whom were clinically depressed. Using 100 photos culled from the workers’ Instagram pages, the researchers evaluated the pictures, gathering data on color, hue, saturation, the number of people in each photograph and the number of likes each picture received.
Reece and Danforth then fed this information to a computer, which used an algorithm to glean the following insights:
Color is a big indicator of whether or not someone is suffering from depression. On Instagram, depressive people tend to gravitate more towards dark colors like blue or gray when posting photos. Depressive users generally prefer to use filters like Inkwell, which transforms colored images into black and white photos, whereas healthier users lean more towards using colorful filters, like Valencia.
Depressive Instagram users don’t always get the love they deserve — their photos generally receive fewer likes than the photos of their healthier contemporaries.
Instagram users suffering from depression post more photos with people in them, but on the whole, there are fewer faces per photo. Reece and Danforth don’t have a scientific explanation for this, but posit that these users might be focusing inward, and that these self-portraits or “sad selfies” might be an extension of the self-focused language that is often attributed to depressed people.
So how well did the computer do in predicting depression amongst ‘grammers? Well, it did a lot better than most doctors: Of the 100 individuals the computer evaluated, it was able to correctly identify 70% of people suffering from depression.
They say a picture is worth a thousand words, and maybe that’s true. But if that picture can also be used to help save a life, than that picture is priceless.
