A MULTI-DIMENSIONAL FRAMEWORK FOR MENTAL HEALTH PREDICTION USING BEHAVIORAL, PSYCHOLOGICAL AND SOCIAL NETWORK DATA

Abstract

Mental health is a vital component of overall well-being, particularly among university students who face diverse academic and social challenges. Conventional methods for stress and mental health assessment often depend on a single type of data, limiting their ability to capture the complexity of human behavior. This study proposes a multi-dimensional framework to improve the accuracy of mental health and stress prediction. The framework integrates heterogeneous data sources, including behavioral patterns, psychological attributes,         social interactions, and environmental factors. Data is collected from mobile usage records, transactional data, self-reported surveys, and online platforms to extract both static features and temporal behavioral trends. A neural network-based model is utilized to identify complex relationships among variables, while a complementary statistical approach ensures scalability and computational efficiency. Temporal analysis is incorporated to track behavioral changes over time, enabling better understanding of evolving mental states. Experimental results show that combining multiple data sources significantly enhances prediction performance. Key indicators include behavioral consistency, social engagement, and interaction patterns. The proposed system provides a scalable and non-invasive solution for early detection and personalized mental health monitoring.