A team of researchers with the University of Vermont and Harvard have published a new study detailing Instagram profiles and the hidden clues they may hold about the photographer’s mental state. Using machine learning, the team was able to identify signs of depression based off an Instagram profile’s photos, metadata, and things like facial recognition. The study looked at 43,950 photographs from 166 individual Instagram users, and had a 70-percent accuracy rate when identifying users with clinical depression.
The artificial intelligence system ultimately proved more capable of detecting depression than general practitioners, which have been found to have somewhere around a 42% accuracy rate. Hints about the photographer’s mental state lie in many things the researchers refer to as ‘markers’: the type of lighting used in the photographs, for example, and the colors of filters applied to photos.
Dark and gray colors are often signs of depression, as well as gap in posting frequency which may indicate a depressed mental state. The number of times a photo is ‘liked’ and commented on, as well as the number of faces detected in the photos, are also notable markers. Interestingly enough, the study found that depressed Instagram users are less likely to use any photo filter, but if they do, they tend to go with ‘Inkwell.’
Via: Digital Trends