Committee: Drs. Amit Sheth, Advisor, TK Prasad, and Tanvi Banerjee
"According to the World Health Organization, more than 300M people suffer from Major Depressive Disorder (MDD) worldwide. PHQ-9 is used to diagnose MDD clinically, and its severity identification. With the unprecedented growth of social media like Twitter, a large number of people have come to share their feelings and emotions on it. These social media messages, specifically tweets can be classified into PHQ-9 symptoms. The current approaches classify tweets to PHQ-9 symptoms do not consider text semantics which is crucial for this classification. In this study, we explore the potential of using Twitter to detect the depression-indicative symptoms such as depressed mood, suicidal thoughts. We provide a semantically enhanced approach to achieve this by using machine learning techniques. Our 2-stage (binary class - multi-label) classification model outperformed state-of-the-art for depression-indicative symptoms classification model by 20%. We further evaluated our semantically-enhanced approach to fill out the PHQ-9 questionnaire and identify the degree of depression from it using the standard guidelines. We show our approach outperforms the existing approaches by examples."