Deep Learning for Early Detection of Depression in Reddit

Abstract

Mental health problems represent a major public heath challenge worldwide, and depression in particular is a serious and widespread form of mental illness. The major issue with depression is that if it is not diagnosed and treated as early as possible, the negative impact it can have on the individual’s life is huge. In many cases, professionals do not have access to patients during early stages of depression unless patients themselves or their relatives report the symptoms to a clinician. However, with the continuous increase in popularity of social media platforms among people all over the world, it was discovered that social media posts could be favorable in detecting mental health illnesses. Therefore, in this study we developed a state-of-art deep learning model to predict the likelihood of a social media user to be diagnosed with depression at an early stage. Then we evaluated the model using an Early Risk Detection Error (ERDE) metric which rewards early detections and penalize late detections. The results show that our deep learning model performs reasonably well on Reddit platform, but there’s more room for improvement, particularly with imbalanced datasets.