Performance analysis of Machine Learning techniques for classification of stress levels using PPG signals

Psychological stress has adverse effects on our autonomic nervous system. The main aim of this research work is to compare the performance and accuracy of various Machine Learning (ML) models used for classification of subjects as having average stress or high stress. Photoplethysmography (PPG) signals is recorded for all subjects after they are introduced to stress inducing stimulus. From the PPG signals Pulse Rate Variability (PRV) parameters are calculated and on the basis of these PRV parameters the subject is either classified as having average stress or high stress. A dataset is framed from the PRV parameters of the subjects and ML models namely Logistic Regression, Support Vector Machine (SVM), Decision Tree and Random Forest are trained and tested, for classification of subjects as average stress or high stress. The models are evaluated based on performance parameters such as confusion matrix, accuracy score, area under curve of Receiver Operating Characteristic (AUC-ROC) of the model, Mean Square Error of the model and other performance parameters. These ML models with good prediction accuracy can be used as an aid by the doctors for more accurate detection of subjects having high stress and thus, begin with the treatment at the earliest.