Andrew G. Reece (Harvard University) and Christopher M. Danforth (University of Vermont) have discovered an algorithm that can identify and detect signs of depression through Instagram feeds.
Data such as colour analysis, metadata components, and algorithmic face detection was collected from over a hundred participants’ Instagram photos.
Current studies of using social media to detect harmful health conditions are through analysing of text. What this duo introduced is using the visual aspect of social media. With over 400 million users, Instagram is one of the dominating social networks. The pair had used this method to analyse 43,950 photos from 166 individuals.
Based on the results from the data, they obtained a higher accuracy rate of diagnosing depression as compared to general practitioners.
What they found was that
- Depressed individuals were more likely to post photos that were bluer, greyer, and darker.
- The more comments the post receives, the more likely it was posted by depressed users, but the findings were opposite for when it came to the number of likes received.
- Depressed users were more likely to post photos with faces, but had an overall lower face count per photo as compared to users without depression.
- Depressed individuals were less likely to apply filters on their photographs.
However, though the results of the study seem very promising, the researchers noted that this is not a definitive measure to accurately diagnose depression. These findings are new approaches which can be used for early screening and detection of mental illnesses.