Detection and Classification of mental illnesses on social media using\n RoBERTa

Abstract

Given the current social distancing regulations across the world, social\nmedia has become the primary mode of communication for most people. This has\nresulted in the isolation of many people suffering from mental illnesses who\nare unable to receive assistance in person. They have increasingly turned to\nsocial media to express themselves and to look for guidance in dealing with\ntheir illnesses. Keeping this in mind, we propose a solution to detect and\nclassify mental illness posts on social media thereby enabling users to seek\nappropriate help. In this work, we detect and classify five prominent kinds of\nmental illnesses: depression, anxiety, bipolar disorder, ADHD and PTSD by\nanalyzing unstructured user data on social media platforms. In addition, we are\nsharing a new high-quality dataset to drive research on this topic. We believe\nthat our work is the first multi-class model that uses a Transformer-based\narchitecture such as RoBERTa to analyze people's emotions and psychology. We\nalso demonstrate how we stress-test our model using behavioral testing. With\nthis research, we hope to be able to contribute to the public health system by\nautomating some of the detection and classification process.\n