Chester Bennington and Chris Cornell committing suicide within months of each other. The problem with depression and the fact that it can happen to anyone, plus that it often goes overlooked, has created a new stir all over again.
It seems the state of a person’s mental health is always obscured by a blur. But a research has shown it is not the case always.
The photos you share online speak volumes.
For some it can be a mode of self-expression, for others a travel-chronicle. They can express quirks of your personality that can go unnoticed by many. They can also be a marker for depression in undiagnosed patients, the research suggests.
Instagram users with a history of depression, for example, from the colour and faces in their photographs, to the various enhancements they use, present their perspective of the world much differently than their peers, according to a study published last week by the journal EPJ Data Science.
Andrew Reece, a postdoctoral student at Harvard University and co-author of the study with Christopher Danworth, a professor at the University of Vermont said, “People in our sample who were depressed tended to post photos that on a pixel-by-pixel basis were bluer, darker and greyer on average than healthy people.”
The duality of “depressed” and “healthy” was established by the pair on the basis of the patients having reported depression (or haven’t) in the past. Machine-learning tools were then used to analyse and find patterns in the photographs to create an algorithmic model for predicting depression by the posts.
Filters or enhancement devices provided by the app itself are instrumental in this study. The filters people use or do not use become a way to study the patterns in people’s mental health.
The pair found how depressed participants used the filter “Inkwell” more often; it drains the colour of the post and renders it black and white. The healthy participants on the other hand used “Valencia” more frequently, which enhanced colours and brightened the post considerably.
The former category, were more than often likely to post photographs with a face. The latter on the other hand, when they post pictures with faces, were likely to post ones containing more faces than one on average.
The pair is optimistic about the promise their technique hold even though the study was exclusive to Instagram.
The group was not just an ordinary one. “This is only a few hundred people and they’re pretty special”, says Mr. Danforth on the study participants. “There’s a sieve we send them through.”
The criteria the group had to go through included being active on Amazon’s Mechanical Turk platform, a paid crowd-sourcing platform, being active on Instagram and being willing to share their entire posting history with the researchers, and sharing the fact that they reported/did not report being depressed in the past.
Out of hundreds of responses, the pair boiled it down to 166 people, 71 of them with a history of depression. They collected nearly 44,000 photographs in all.
The researchers then used software to analyse each photograph’s tint, colour saturation, brightness and the number of enhancements used (if used at all). Also, they studied the number of faces in the photograph if there were any. Comments and likes on each post were also part of the study.
Using machine-learning tools, Mr. Reece and Mr. Danforth found that the more comments a post received, the more likely it was to have been posted by a depressed participant. The opposite was true for likes. And depressed users tended to post more frequently, they found.
Though warning that these patterns weren’t applicable to all users, the pair is hopeful about similar models being applied in mental health screenings someday.
“We reveal a great deal about our behaviour with our activities,” Danforth said, “and we’re a lot more predictable than we’d like to think.”