Modern computer algorithms are capable of much today. They are able to drive a car, select the most optimal playlist according to the preferences of the user of streaming services like Spotify, can retouch photos and even describe the surrounding world to blind people. And now, machine learning is able to look into your Instagram service tape, analyze it and decide in what psychological state you are and whether you have clinical depression.

The results of the last study published on the SpringerOpen website speak about the analysis of 166 bands of Instagram users, within the framework of which the machine was looking for so-called "markers of clinical depression". The authors of the study argue that the evaluation carried out by the computer algorithm was even more successful than the cases of detection and diagnosis of this condition by real practitioners.

"The statistical data was collected on the basis of an analysis of a total of 43,950 photographs of Instagram users who evaluated the color gamut, metadata, and also used the algorithmic method of face recognition. The results of the effectiveness of identifying signs of depression have surpassed the average performance indicators of general practitioners ".

The algorithm was tested immediately by a number of different factors, ranging from the color gamut of the photograph (including checking the use or non-use by the user of various color filters) and ending with the frequency of publications.

"In studies of mood, color and mental health, it was revealed that the darker and more gray-colored psychologically healthy people are most often associated with a bad mood, and the brighter and more saturated with the good. Our examination showed that people in a depressed state choose darker and gray tones for their photos [...] ".

"Depression is closely related to decreased social activity. Since Instagram is used for the exchange of personal experience, it would be reasonable to conclude that published photos can also reflect the social aspects of a user's life. Based on this, we used the algorithm for determining individuals for the analysis of publications in Instagram to determine the presence and number of human faces on each of the photographs. We also calculated the number of comments and likes under each post, reflecting community involvement, and used the frequency of publication as a measure of user engagement ".