News
Your Instagram photos may show if you are depressed
Depression may be one of the most widely-recognized mental health issues facing the public, but estimates suggest that up to two-thirds of people with depression never seek or receive treatment.
The problem is that currently all diagnosis methods require the patient to willingly speak up about their struggles or the health professional to be able to proactively identify warning signs. Considering many people with depression don’t even realize it and may not be willing to speak openly about their symptoms, the lack of proactive ways to identify depression make it difficult to more widely address the mental illness.
To help address this dearth of methods to pre-emptively identify depression, a team of researchers at Harvard University and the University of Vermont set out in search of a way to spot the mental illness on a wide scale and stumbled on a potential lead in an unsuspected place – Instagram.
According to the small study published in the EPJ Data Science Journal, machine-learning algorithms could detect signs of depression by evaluating images shared on the popular social media platform. Specifically, the algorithms analyzed factors like how many likes and comments a photo received, the number of faces in an image, and the colors in the pictures. It also compared whether the individuals used filters on their images.
The team then compared the findings of the system against the answers given by participants on the Center for Epidemiologic Studies Depression Scale questionnaire – a widely used tool for diagnosing depression.
Based on an evaluation of nearly 44,000 photographs posted by 166 study participants (71 of which had a history of depression), the researchers found that depressed users tended to share photos that were bluer, darker, or greyer, received less likes. The pictures were also less likely to have filters used on them. When the depressed individuals did use filters, they typically opted for black and white color schemes.
While the findings are very interesting, even the researchers who led the study were quick to emphasize that the experiment was very small and shouldn’t be seen as hard proof of the concept. Instead, they say the results validate the potential “promise of their techniques” and require further study.
It is also important to note that while systems like these may one day be useful for identifying those who are potentially depressed, they will never replace a proper in-person consultation and assessment.
