Instagram Photos Could Help Diagnose Mental Illnesses
New research shows that the pictures people post to their social media accounts can indicate their mental and emotional states
Mental illness can be difficult to diagnose, and even harder to detect. Researchers are hoping to use social media to better find and help those in need, and a new report says that pictures posted to Instagram can point to potential. In a study published in arXiv, Harvard Ph.D candidate Andrew G. Reece and University of Vermont professor Christopher M. Danforth looked at 170 Amazon Mechanical Turk workers, 70 of whom were clinically depressed at the time. These workers completed a survey, and from there, their Instagram accounts were analyzed by an algorithm looking for patterns for those who were not depressed as well as photos taken before the diagnosis in the case of those who were depressed.
100 photos were analyzed per user, and people rated them on a scale of 0 to 5 for happy, sad, or how interesting they were. The photos were also categorized by saturation, hue and the number of faces in each. The results indicated that darker color schemes meant that a person was more likely to be depressed, especially those who used grayscale filters. Those who were not depressed were more likely to use highlighting filters like Valencia.
They were able to indicate which users were clinically depressed with a 70 percent success rate. The researchers ultimately conclude:
“More generally, these findings support the notion that major changes in individual psychology are transmitted in social media use, and can be identified via computational methods.”
While this is a promising start to finding new ways to help those who need therapy and other help, it is not going to be implemented any time soon. Social media is not an accurate representation of a person’s life, as they can choose what they do and don’t share. Users can also use social media to present a specific kind of image, meaning that people can present themselves as happier than they actually are.