Early risk detection of depression from social media posts using Hierarchical Attention Networks

Abstract

Efficient mental health diagnosis is improving continuously, yet many cases go undetected. Early detection of depression can potentially prevent people from mental illness and live a better life. There are many ways to monitor depression in people; the most obvious one is to monitor the messages posted by people on social media platforms. In recent years, detecting early depression from social media posts has been a focused research area. In this paper, we use Hierarchical Attention Networks, a Deep Learning-based method to classify whether the users are depressed or not using their historical, social media posts.