Instagram photos reveal predictive markers of depression

0425

Hypothesis 1 Instagram posts made by individuals diagnosed with depression can be reliably distinguished from posts made by healthy controls, using only measures extracted computationally from posted photos and associated metadata.

Hypothesis 2 Instagram posts made by depressed individuals prior to the date of first clinical diagnosis can be reliably distinguished from posts made by healthy controls.

Hypothesis 3a Human ratings of Instagram posts on common semantic categories can distinguish between posts made by depressed and healthy individuals.

Hypothesis 3b Human ratings are positively correlated with computationally extracted features. (showed little or no correlation with most computational features)

Method

1. User activity (number of posts)
2. Community reaction (like it)
3. Human face (number of faces, ~)
4. HSV (Hue, Saturation, and value)

Post , 1/ like it , 1/number of faces, 1/ Filtered~ Depressed

The more comments Instagram posts received, the more likely they were posted by depressed participants, but the opposite was true for likes re-
ceived.

Depressed participants were more likely to post photos with faces, but had a lower average face count per photograph than healthy participants.

Figure 1 Comparison of HSV values. Right photograph has higher Hue (bluer), lower Saturation (grayer), and lower Brightness (darker) than left photograph. Instagram photos posted by depressed individuals had HSV values shifted towards those in the right photograph, compared with photos posted by healthy
individuals.

Figure 2 Magnitude and direction of regression coefficients in All-data (N = 24,713) and Pre-diagnosis (N = 18,513) models. X-axis values represent the adjustment in odds of an observation belonging to the target class, per unit increase of each predictive variable. Odds were generated by exponentiating logistic regression log-odds coefficients.

Figure 3 Instagram filter usage among depressed and healthy participants. Bars indicate difference between observed and expected usage frequencies, based on a Chi-squared analysis of independence. Blue bars indicate disproportionate use of a filter by depressed compared to healthy participants, orange bars
indicate the reverse. All-data results are displayed, see Additional file 1 for Pre-diagnosis plot.