Social Media Based Depression Intensity Prediction using NLP & DL

Mental health disorders, particularly depression, pose a significant public health challenge, affecting over 280 million people globally and ranking as the leading cause of disability worldwide. Traditional assessment methods like the Patient Health Questionnaire-9 (PHQ-9) rely on clinical interviews and questionnaires, which are subjective and may not fully capture the dynamic nature of symptoms, often leading to under-diagnosis and under-treatment. To address these limitations, the Depression Intensity Analyzer (DIA) project was developed as a comprehensive web-based application that leverages state-of-the-art Natural Language Processing (NLP) and Deep learning techniques to automatically assess depression severity from textual input. The system’s core innovation is its dual-mode operation, which allows for automatic emotion detection using a pre-trained transformer model or manual emotion input through sliders. At its foundation, the system employs a dual-output Keras deep learning model that simultaneously predicts a continuous depression intensity score (0-4) and classifies the text into discrete severity levels, providing both granular and categorical insights. Built on the Flask web framework, the DIA system is designed for real-time processing, providing an accessible, scalable, and objective tool to complement traditional mental health assessment and monitoring.