ECG Stress Detection Leveraging Deep Learning Techniques

Abstract

Accurate stress detection is crucial in modern healthcare, as prolonged stress contributes to a range of physical and mental health conditions. This paper introduces a deep learning-driven framework for classifying stress using electrocardiogram (ECG) signals, aiming to enhance detection accuracy through advanced neural network models. The study utilizes the MIT-BIH Arrhythmia Database, applying a standardized pre-processing pipeline before training and evaluating three neural architectures: Gated Recurrent Units (GRU), Long Short-Term Memory (LSTM), and Convolutional Neural Networks (CNN). Performance was assessed using key metrics including precision, recall, F1-score, and the area under the receiver operating characteristic (ROC) curve. Experimental results indicate that the GRU model delivered superior performance, achieving a peak accuracy of 98%, surpassing both LSTM and CNN models. These findings demonstrate the effectiveness of deep learning for real-time, ECG-based stress monitoring and underscore its potential for integration into wearable health monitoring technologies.