Stress Detection Based on Human Sleep Cycle

Abstract

Anxiety, mental or emotional stress resulting from a challenging environment, significantly impacts well-being. Understanding and monitoring human stress levels are crucial for averting adverse outcomes. This study investigates stress detection through machine learning and deep learning algorithms, focusing on sleep-related behaviours. Six machine learning techniques and deep learning methods, including Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Deeper Neural Networks, were employed and compared against benchmarks from prior studies. Notably, the Naïve Bayes algorithm exhibited exceptional performance, achieving 91.27% accuracy. The integration of deep learning methods provided a broader perspective on stress detection and complemented insights from established studies. Leveraging previous research results not only served as benchmarks for our model but also validated and extended our understanding of stress detection based on sleep-related behaviours. Our findings contribute to the discourse on human stress monitoring.

Keywords

• stress
• machine learning
• deep learning
• sleep
• naïve bayes
• accuracy
• human monitoring