The candid pictures on Instagram may give followers a view into the state of the account-holder's mental health, a study has revealed.
Researchers Andrew Reece from Harvard and Chris Danforth from the University of Vermont issued a report this week that reveals how social media can be used as a tool in early detection and intervention in psychiatric disorders.
The study looked at over 43 000 photos from Instagram and applied artificial intelligence (AI), using colour analysis, metadata components and algorithmic face detection, to identify the visual markers of depression.
The thousands of photos came from 166 Instagram accounts of people who were both mentally healthy and clinically depressed.
The model was able to indicate with 70% accuracy who was diagnosed as depressed and who was not.
Using only photographic details, it "performed better than the rate at which unassisted general practitioners typically perform during in-person patient assessments", states the report.
The blues
Hue, saturation and brightness levels were looked at and compared in photos from the social network.
Results showed depressed individuals tend to post pictures with higher hue and lower saturation and brightness points. This means their pictures were more likely to be bluer, greyer and darker.
However, they are less likely to post pictures with filters applied. When they did use them, they were more likely to use the 'Inkwell' filter, while their mentally healthier counterparts favoured 'Valencia'.
Other factors include the number of likes and comments on a photo. More comments and fewer likes were seen on photos uploaded by those depressed.
The report also shows depressed people post pictures of faces more frequently, but had a lower average face count per photo than healthier individuals.
These factors could all indicate depression, but researchers say they should not be taken as enduring facts, "rather as a methodological foundation upon which to build and refine subsequent models", concludes the report.
"There is good reason to prioritise research into Instagram analysis for health screening," say the researchers.
"Instagram members currently contribute almost 100 million new posts per day, and Instagram's rate of new users joining recently outpaced Twitter, YouTube, LinkedIn and even Facebook."
Share