Fuzzy Logic and Deep Learning Hybrid Model for Physiological Stress Detection in Higher Education

The identification of stress in higher education students provides a significant challenge that impacts academic performance and mental health. Conventional stress classification methods show low accuracy and inadequate interpretability due to the complexities of physiological data fluctuations. This study presents a hybrid model integrating deep learning and fuzzy logic, which integrates Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and a fuzzy inference system to improve accuracy and interpretability. The model was trained using physiological datasets, including heart rate variability, electrodermal activity, and EEG signals. Experimental findings indicate an important advance in stress classification, attaining 98.7% accuracy, 97.8% precision, 98.1% recall, and an AUCROC score of 0.995. The suggested approach demonstrates enhanced performance in differentiating stress levels compared to baseline models such as SVM and Random Forest, while also providing human-interpretable insights. This study offers a detailed method for educational institutions to monitor student stress in real time, facilitating early interventions and improved academic results. Future works may focus on practical implementation and adaptive fuzzy systems to enhance realtime stress evaluation and tailored interventions.