Short of someone holding up a sign and saying that he or she is depressed, can Instagram postings really help diagnose depression? A recent study suggests possibly, but be careful how much you read into this study...and Instagram pictures.
Here's what Andrew G. Reece from Harvard University and Christopher M. Danforth from the University of Vermont did for a study published in EPJ Data Science. They used Amazon’s Mechanical Turk (MTurk) crowdwork platform to eventually recruit 166 individuals, who agreed to share their Instagram data and whether they already had a clinical diagnosis of depression (71 had a history of depression). Each study subject took the CES-D (Center for Epidemiologic Studies Depression Scale) questionnaire to identify which participants may have depression. The researchers then used computer algorithms to sift through and categorize over 43,000 photos to determine what characteristics may be associated with depression as determined by the questionnaire.
People who were depressed were more likely to:
But before you get worried about those photos that you posted with only half your face sticking out of a blueberry costume that got a hundred "what is that?" comments but few likes, keep in mind that this is a limited study. It's difficult to draw strong conclusions from a relatively small and select sample of people. Moreover, there may be many other reasons for posting certain types of photos. For example, if you are a rabid Seattle Seahawks fan, many of your photos may be blue and gray. Similarly, if you actually play for the Seattle Seahawks, then your football helmet may be frequently obscuring part of your face. Moreover, if frequency of posting is a criteria, then what does this say every celebrity very active on Instagram...although they all probably get a lot of "likes."
The researchers also make a rather large leap when they say that "our model showed considerable improvement over the ability of unassisted general practitioners to correctly diagnose depression." What the study compared is how well their computer algorithms was able to detect depression as determined by the questionnaire (70 percent of the time) with what percentage of those identified by the questionnaire as depressed actually had a self-reported history of depression (42 percent). Remember to get a diagnosis of depression, you first have to go to a health professional who is qualified to make such a diagnosis (e.g., a podiatrist won't be looking for this) and then you have to reveal to the health professional your symptoms. Many people who are depressed may not even regularly see a doctor. Thus, the study really didn't compare their algorithm with the abilities of a properly-trained health professional.
While this study shows some interesting trends, it by no means shows a viable way of diagnosing depression. You can't diagnose depression from afar. This is true with most mental health issues with some exceptions that are usually extreme cases. For example, Alien Hand Syndrome (your hand has a life of its own) or Walking Corpse Syndrome (you believe that you are dead, not a Grateful Dead fan or really tired, but actually a zombie), and Self-Cannibalism (which is self-explanatory) may be easier to diagnose. However, these days people throw around what sound like mental health diagnoses without any proper professional evaluation...such as rivals calling each other "nuts", "crazy", "insane", "depressed", "paranoid", "narcissistic" or other things that sound like mental health labels.
Another danger is stereotyping. Our society already relies too heavily on superficial appearance to draw grand conclusions. Judging someone based on their appearance alone is laziness, and the risk is that this will extend to Instagram pictures. Rather than taking time to really get to know people before making judgments, might you say "she doesn't have enough likes for me" or "too many black-and-white photos, something must be going on" or "hmm, that pet pig keeps covering part of his face"?
Of course, studies have shown that depression is both under-diagnosed and under-treated, meaning that there needs to be more ways of identifying those who could use help. This stems from several issues such as:
How can Instagram possibly help? It's hard to say. You probably don't want to start sifting though everyone's Instagram posts to determine who may be depressed. And I don't quite see a doctor saying "take off your shirt, take a deep breath, and show me your Instagram posts" or "I've been following you on Instagram lately. What's up with the lack of filters?" Routinely diagnosing depression solely or largely on Instagram posts just should not happen.