A new study suggests Instagram photos could provide precious clues to whether you have depression or not. Researchers were able to craft an algorithm based on some factors that appear in those pictures to estimate if a person should be diagnosed with a mental health issue.
The small study which involved 166 people and 43,000 Instagram photos included questions on every volunteer’s mental health history. Around half of participants had had a diagnosis of depression in the last three years.
The new algorithm takes into account a number of factors such as the number of people in a photo, the number of likes, and colors used in the snapshot. Instagrammers that used darker hues and fewer people in their photos were more likely to suffer from depression.
Those with depression were also not likely to use filters on their photos, but when they did opt for a filter they used ‘Inkwell’ which turns color photos into black and white ones. Users with no mental health users were more likely to use the vibrant colors of the ‘Valencia’ filter.
The new algorithm researchers crafted determined if an Instagrammer had depression with a 70 percent accuracy. The software is able to pinpoint clues that the person might have depression even before an official diagnosis.
New Research Not 100% Fail-Proof
The recent research is just a proof of concept, which means that its authors just wanted to know if a theory or software can have a real-world application. So, it shouldn’t be seen as 100% correct yet.
Furthermore, the study had its limitations as the sample size was relatively small. All study participants agreed to answer questions about their mental health status and were relatively active on the photo-sharing site. So, it is unclear if the findings can apply to an average social media user.
The research team, though, is confident the study can encourage other researchers to use technology to diagnose depression in Internet users. If the efforts yield positive results, depression could be detected earlier with better treatment outcomes.