DUDetector: Detecting Depressed Users in Twitter Using Deep Learning

Many individuals use social media platforms to discuss their mental health issues and feelings, providing a rich source of data for analysis. Therefore, social media platforms can be a valuable source of data for detecting depression. Early detection and intervention can greatly improve outcomes for individuals suffering from depression. Furthermore, analyzing social media data for depression detection can provide a more efficient and scalable solution compared to traditional methods. In this paper, we propose a novel framework, called DUDetector, for detecting depressed Twitter users using deep learning. The proposed DUDetector approach uses two sources of information: the user's combined tweets and their “about” section. To identify depression in Twitter users, DUDetector utilizes Bidirectional Encoder Representation from Transformers (BERT) embeddings and Bidirectional Gated Recurrent Unit (BiGRU) models. The collected data is fine-grained to extract relevant content related to depression by using word frequency, which improved the performance of our model. Experimental results reveal that the proposed DUDetector framework outperformed all existing solutions for detecting depressed users on Twitter.