INSIGHT: A next-generation framework for proactive mental health detection using advanced data fusion and deep learning

University students currently undergoing physical and mental development experience serious mental health obstacles that can result in depression and self-inflicted injury. Medical help does not reach students who do not know that they are affected by mental health disorders because early diagnosis requires precise detection of mental conditions. Campus environments produce diverse multi-modal unstructured data that makes mental health identification difficult by its nature. Our framework INSIGHT (Intelligent Student Mental Health Detection Framework) provides a novel solution which applies advanced methodologies to improve mental health detection accuracy. This framework unifies three distinctive advanced elements. A multi-modal data fusion strategy uses Graph Neural Networks (GNNs) to integrate social behavioral, academic pattern and physical activity data to create an all-encompassing student life model. The proposed adaptive generative data augmentation method (ADAM) conducts dynamic synthesis of high-quality minority samples to enhance model robustness through label imbalance mitigation. The final model uses Attention-based Long Short-Term Memory Networks (ALSTM) together with Transformer encoders to achieve precise mental health prediction outcomes. The INSIGHT solution demonstrates superior performance according to extensive experiments that achieve precision gains of up to 94.7% compared to base methods. University mental health prevention detection through the proposed system serves as an effective predictive tool which delivers actionable insights to improve campus mental health support efforts.