MULTI CLASS STRESS DETECTION THROUGH HEART RATE VARIABILITY A DEEP NEURAL NETWORK BASED STUDY
DOI:
https://doi.org/10.64751/Keywords:
Stress detection, heart rate variability (HRV), deep learning, convolutional neural network (CNN), multi-class classification, physiological signals, SWELL-KW dataset, ANOVA, feature extraction, time-domain features, frequency-domain features, mental health monitoring, stress classification.Abstract
The study titled "Multi-Class Stress Detection Through Heart Rate Variability: A Deep Neural NetworkBased Study" explores stress detection using heart rate variability (HRV) as a physiological biomarker. Stress, a natural human response to pressure, can become chronic and lead to mental health issues like anxiety, depression, and sleep disorders. Traditional stress measurement methods based on HRV face challenges in achieving high accuracy. Unlike heart rate, which measures the average beats per minute, HRV represents the variation in time intervals between consecutive heartbeats, specifically the RR intervals. This study focuses on leveraging HRV features for multi-class stress detection by developing a Convolutional Neural Network (CNN)-based model. The model classifies stress into three categories: no stress, interruption stress, and time pressure stress, utilizing both time- and frequency-domain HRV features. The research is validated using the SWELL- KW dataset, achieving an impressive accuracy of 99.9% (Precision=1, Recall=1, F1-score=1, and MCC=0.99), surpassing existing methods. Additionally, the study highlights the significance of HRV features in stress detection through a feature extraction technique based on analysis of variance (ANOVA), demonstrating the effectiveness of deep learning in stress classification.
