Detection of Depression and Suicidal Tendency Using Twitter Posts

Abstract

It was established that between 1987 and 2007, the suicide rate burgeoned from 7.9 to 10.3 per 100,000, with higher suicide rates in southern and eastern states of India. India does not only face the fear of suicide but also of depression. A study reported in the World Health Organization (WHO), conducted for the National Care Of Medical Health (NCMH), states that at least 6.5% of the Indian population suffers from some form of the serious mental disorder, with no discernible rural–urban differences. The key challenge of suicide and depression prevention is understanding and detecting the complex risk factors and warning signs that may precipitate the event. In this project, we present an approach that uses the social media platform to quantify suicide-warning signs for individuals, to evaluate a person’s mental health and to detect posts containing suicide-related content. The pivot point of this approach is the automatic identification of sudden changes in a user’s online behaviour. To detect such changes, we combine natural language processing techniques to aggregate behavioural and textual features and pass these features through a martingale framework, which is widely used for change detection in data streams.