A picture speaks a thousand words, but our Instagram photos could say a lot more about our mental health than we realize.

In fact, researchers from Harvard and Vermont University created a computer algorithm that analyzed Instagram data to help identify markers of depression. After looking through 43,950 photos from 166 people (71 of whom had received a depression diagnosis), the algorithm was able to correctly identify participants with depression 70 percent of the time.

That’s a statistic that’s higher than general practitioners, who have about a 42 to 50 percent accuracy.

So how did the algorithm look at the data?

The study, published in the journal EPJ Data Science, looked at 43,950 photos from the participants and analyzed the photos by color spectrum:

  • hue, or how red or blue to purple the coloring of the photo is
  • saturation, or how vivid an image is
  • value, or how bright the image is

People who’ve received a diagnosis of depression were more likely to post photos following a certain pattern:

Pattern from other studies on depression or moodInstagram posting pattern from study
more likely drawn to blue and gray colorsbluer, grayer, and darker pictures
may have reduced interactions with peopleface photos had fewer faces per photo
black, white, and gray may be associated with less positive moodsmore likely to use black and white filters

Other seemingly obvious indicators, like the number of likes and comments, fell flat upon the test, not predicting much about the participants’ mental health.

Can we really predict depression based on Instagram photos?

One of the authors of the study, Dr. Chris Danforth, stressed to Buzzfeed News that the algorithm isn’t a diagnostic test. Instead, it could potentially become a tool to help give warnings or indications of when a person is showing signs of depression.

For example, when the algorithm was also given photos it had never seen before — photos from before some participants’ clinical diagnosis — the algorithm was able to pick up signs. It also still outperformed general practitioners under these conditions.

One thing’s for sure. The results of the study shouldn’t make any person feel like an expert on depression.

For starters, the sample size of this study is too small to attribute these findings to the average Instagrammer.

Second, the study doesn’t tell us the range of blue to gray the algorithm looked at. A color wheel study found that blue was, on average, the favorite color in all participants, whether they were healthy, anxious, or depressed.

Last, only a finely tuned algorithm would be able to look at the photos, from pixel to pixel, without bias.

The future of apps, social media, and therapy

The relationship between social media and mental health has always been a bit of a roller coaster. Some studies report that prolonged use of Facebook has correlations with depression while others present opposite research, finding a positive impact on self-esteem.

The results of this study, however, do point toward a more hopeful, engaged future between people and their doctors or therapists.

“Imagine an app you can install on your phone that pings your doctor for a checkup when your behavior changes for the worse, potentially before you even realize there is a problem,” Dr. Danforth told EurekAlert! in a press release.

But before this app ever comes out, we have to fix the bridge of trust between therapy and users. Out of the 509 participants originally recruited, 43 percent dropped out because they didn’t want to share their Instagram data. And even if they did, it doesn’t mean they would necessary be honest with their therapist. Another 2016 study published in the Counselling Psychology Quarterly found that 72.6 percent of 547 adults reported lying to their therapist about at least one topic.

The study notes that “the more comments Instagram posts received, the more likely they were posted by depressed participants.” But it doesn’t go into detail about how many followers these Instagrammers had, who posted the comments, and whether the poster themselves were interested in engaging in conversations.

However, it’s nice to imagine that these comments were made in hopes of discussion — an open discussion about mental health where judging doesn’t start from a photo or someone’s favorite color.

Christal Yuen is an editor at Healthline.com. When she’s not editing or writing, she’s spending time with her cat-dog, going to concerts, and posting gray, unsaturated photos on Instagram. You can reach her on Twitter and Instagram.