Student Anxiety Prediction Using Active and Deep Learning Techniques

Mental health disorders, such as anxiety, are often diagnosed subjectively, leading to inconsistencies and delays in treatment. With the growing availability of physiological signal data, there is an immediate demand exists for automated solutions capable of detect anxiety patterns effectively. Early detection and continuous monitoring of student anxiety are essential for effective support and intervention. However, traditional approaches, while helpful, often struggle with accuracy and efficiency due to their limited sensitivity and specificity. Student anxiety forecasting has mainly depended on machine learning models like Random Forest, XGBoost and Decision Tree. This research introduces an active learning- based deep learning framework for anxiety detection. These strategies provides moderate accuracy, difficult to handle the large datasets. To resolve these limitations in student anxiety prediction, Recurrent Neural Network(RNN)and more enhanced variants like LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Units), and Active Learning will be deployed. An active learning loop ensures optimal model performance by iteratively retraining on samples with high uncertainty, significantly reducing the dependency on large labeled datasets. These advancements position the research as a viable candidate for real-world applications in mental health diagnostics.