Depression Symptoms Identification through Social Media Data: Applying Design Science Research to Develop a Classification Model

The thesis addresses the prevalence of depression as a debilitating condition and underscores the importance of early identification of symptoms for timely interventions. Exploring user-generated content on social networks, the study proposes the use of machine learning models in detecting depressive symptoms. Following the DSR methodology, the research validates the effectiveness of these models compared to existing approaches, involving healthcare professionals and domain experts. The thesis introduces an innovative stacking model using LIWC metrics from social media posts, contributing to the understanding of machine learning-based solutions in identifying symptoms of depressive disorder.