Andrew Reece of Harvard University and Chris Danforth of the University of Vermont collected the data of 166 Instagram users — a total of 43,950 photos, and their mental health history. Half of the participants had been diagnosed with depression within the past three years.
After analyzing the photos and mental health history, scientists found that they could predict who had depression 70 per cent of the time. The study, published in EPJ Data Science on Monday, also analyzed more than 100 studies on how accurately doctors can diagnose depression.
“General practitioners were able to correctly rule out depression in non-depressed patients 81 per cent of the time, but only diagnosed depressed patients correctly 42 per cent of the time,” the study reads.
Individuals with depression have “different preferences” when it comes to colour, shading, brightness of photographs, the study’s authors wrote in a blog post.
Those who are depressed typically have darker photos, with blue and grey hues. They also found fewer images of faces on their Instagram feeds, which researchers likened to a key characteristic of depressed individuals who tend to socialize less.
The differences between mentally healthy and depressed people were tracked down to the filters on Instagram. Healthy individuals used the filter Valencia most often, while depressed individuals used Inkwell, which is much darker.
Proving a link between social media habits and mental health could change the way diagnosis are made, the blog post notes.
However, the study does highlight some limitations to using social media as an indicator of mental health. There are privacy concerns; some users may not feel comfortable with sharing their accounts for analysis. Technology also evolves quickly, the study notes, which means the method requires frequents checks and updates.
The study offers only “promising leads” into new mental health screening methods, it notes.