Enhancing Human Stress Detection from Optimized and Selected Features Using Feed Forward Neural Network

Stress, a psycho-physiological response to life’s challenges and changes, poses significant health risks when chronic. This study explores the link between stress and sleep quality, visualizing how stress affects the human body. Stress is categorized into five levels, and eight key physiological parameters are analyzed. Machine learning algorithms, including Naïve Bayes, Support Vector Machine, and Random Forest Classifier, achieved an impressive 100% accuracy in stress level classification. Further enhancements using Principal Component Analysis (PCA) and Chi-square tests identified optimal features, resulting in approximately 99.68% accuracy with a feed-forward neural network (Multilayer Perceptron). These findings advance stress detection and its impact on sleep quality, offering potential applications in real-world health monitoring systems. The model’s high accuracy provides valuable insights for stress management and preventive health measures, mitigating associated risks.