A new study conducted by Chris Danforth of the University of Vermont and Andrew Reece of Harvard University reveals that photos posted on social media can reflect a person’s mental status. The pair has created a machine that checks Instagram photos to formulate a connection between usage of color and the mental state of a person. They used machine learning to evaluate more than 43,000 photos from 166 Instagram accounts. The pictures and the filters were thoroughly reviewed to reach a conclusion.
The researchers believe that people posting images carrying a blue or a gray tone on Instagram are more likely to be depressed than those who post images with vibrant color. The color was not the only criteria to study mental health of a person, as data also was collected from a facial recognition algorithm. Both the data was combined to see if machine learning could accurately identify depressions.
How was the study conducted?
For the purpose of the study, researchers examined 170 workers with an Instagram account from Amazon’s Mechanical Turk service. All the workers were to complete a questionnaire which also included a standard clinical depression survey. After the survey, the Turks were asked to share the Instagram post for the study.
Then the machine learning model was able to detect which workers showed signs of depression just by using color analysis, metadata components, and algorithmic face detection. About 70 percent of the time the model accurately predicted the signs of depression of the social media user. The prediction was so high that the researchers believe their technique can be used to detect early evidence of mental illness.
What did the study find out?
Researchers learned that people who tend to share darker, bluer and more faded photos on their feeds are more likely to be depressed. Also, researchers noted that people with depression usually receive fewer likes on their photo in comparison they comment a lot on other people feed.
Further, the pair claimed that though depressed people post may include faces in the picture, but are more inclined to self-focused language. The study also revealed that depressed users tend to use more filters in their photos like Inkwell which changes the color image into black and white. On the other hand, healthy participants mostly use brighter Valencia filter which lightens the tints of the pictures.
In general terms, the study suggests that significant changes in individual psychology get transmitted in social media which can be recognized through the computational methods.