I Need Help, not Likes: Review of Models to Detect Depression through Social Media Data

Mental health is one of the most pressing challenges that we as humans are facing today. Despite the presence of various resources to detect and diagnose mental disorders, it continues to be an issue for a significant proportion of the world. Researchers are now exploring the application of natural language processing and deep learning on the diagnosis of these disorders.However, due to the lack of a systematic and organized set of data to train upon, many of them have often failed. In this review paper, we explore prominent techniques to diagnose a specific mental disorder – Depression. We analyze and review the work done in the field of depression detection from feeds of social media platforms. For our study, we investigated mainly 3 platforms – Facebook, Twitter, and Reddit.In the later sections of the paper, we aim to analyze another pressing topic of our study- The suitability of Social Media sources to act as data resources for the diagnostic tasks for mental health disorders. As the inclination of society towards social media increases, it evolves to become a better and robust outlet for people’s emotions and expressions. Whilst recording and obtaining logs of a particular patient’s visit with the therapist might be a more accurate description, however, it is one that is usually untimely and unlikely to be formally organized. In such scenarios, a patient’s posts on various social media platforms and his activity on the same provide a better turn of perspective to solve such problems.