Classification of Mental Stress Using CNN-LSTM Algorithms with Electrocardiogram Signals

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Artificial Intelligence in Healthcare

Background:

• Mental stress is a significant contributor to chronic diseases like depression, cancer, and cardiovascular conditions.
• Regular monitoring and management of an individual's stress levels are crucial for public health.
• Electrocardiogram (ECG) signals exhibit measurable changes under stress, offering a potential biomarker.

Purpose of the Study:

• To develop and validate a highly accurate ensemble algorithm for identifying mental stress states.
• To leverage a modified Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) architecture for stress detection.
• To enhance stress classification by analyzing ECG signals in both time and frequency domains.

Main Methods:

• Utilized a modified CNN-LSTM architecture for analyzing ECG signals.
• Employed Fast Fourier Transform (FFT) and spectrograms for signal preprocessing, capturing time and frequency domain features.
• Evaluated model performance using confusion matrices, Receiver Operating Characteristic (ROC) curves, and Precision-Recall (PR) curves.

Main Results:

• Achieved a classification accuracy of 98.3% for mental stress detection.
• Demonstrated a significant improvement of 14.7% compared to previous research findings.
• The proposed model effectively classifies stress signals based on ECG modifications.

Conclusions:

• The developed ensemble algorithm provides a robust and accurate method for mental stress monitoring.
• This technology can significantly contribute to proactive mental health management strategies.
• Future integration with other biosignals (EMG, PPG) holds potential for diverse healthcare applications.