EARLY DETECTION OF DEPRESSION VIA SOCIAL MEDIA

The pervasive nature of social media has fundamentally altered the landscape of human communication, leading to a profound reliance on digital platforms for emotional expression. This increased digital footprint provides a critical opportunity for the early detection of depressive states, which have seen a correlated rise with social media usage. This research proposes a computational framework utilizing Natural Language Processing (NLP) and Deep Learning to serve as a Decision Support System (DSS) for counselors, psychologists, and criminal investigators. By employing Word2Vec embedding techniques integrated with a hybrid Convolution Neural Network (CNN) and Long Short-Term Memory (LSTM) architecture, we analyze textual data to identify latent indicators of mental health decline. Our model evaluates the effectiveness of context-free embeddings in classifying sentiment within large-scale social media datasets. The results established in this study provide an empirical baseline for automated mental health monitoring, underscoring the trajectory from static embeddings toward sophisticated contextual modeling.