Study of a Hybrid CNN-SVM Model for Stress Detection with Automated Heart Rate Variability Feature Extraction Method

In the era of big data, stress detection techniques based on electrocardiogram (ECG) data play a crucial role in monitoring mental health and preventing mental disorders. Given that the relationship between various heart rate variability (HRV) parameters extracted from ECG signals and stress levels requires further exploration, this paper presents a hybrid model combining a Convolutional Neural Network (CNN) and a Support Vector Machine (SVM). The proposed model leverages the CNN's robust feature extraction capabilities alongside the SVM's superior performance in high-dimensional classification tasks to achieve accurate stress detection. For this study, two publicly available datasets, WESAD (Wearable Stress and Affect Detection) and SWELL (SWELL-KW), are employed, both containing ECG signals recorded under different stress conditions. Experimental results demonstrate that the CNN-SVM model achieves a stress detection accuracy of 0.99 on both datasets, underscoring its potential for reliable stress detection tasks.