Bipolar disorder detection over social media

Abstract

Bipolar disorder is a mental illness characterized by manic and depressive episodes. The inability to track the patient at different stages of the disease, the patient’s concealment of information, and the difficulty in obtaining and paying for a psychologist are all weaknesses in traditional diagnosis procedures. In this regard, computer researchers have developed automated prediction algorithms in response to issues involved in using traditional ways of diagnosing bipolar disorder. Although these automated approaches have eliminated many problems that plagued previous systems, there are still many challenges to tackle. Discovering a mechanism to track changes in user behavior and aggregating various features into a cohesive model are the most critical issues in this context. To address these concerns, this research proposes a novel approach for detecting bipolar disorder among Twitter users.

Keywords

• Bipolar disorder
• Social media
• Feature engineering
• Twitter
• Machine learning
• Predictive models