Comparative Analysis of Machine Learning Methodologies for HRV based Stress Detection

Stress detection is a critical field in physiological and mental health research, with heart-rate variability (HRV) serving as a key biomarker of stress levels. With good understanding of HRV patterns under stress, quality assurance and intervention follow-ups can be put in place for mental health. This paper examines the performance of the machine learning model in detecting stress using the SWELL-KW dataset, which records physiological responses in diverse stress settings. To enhance performance, a structured approach was adopted with preprocessing techniques that range from standardization through scaling. Feature selection was done with ANOVA, correlation analysis, and variance thresholding, refining the dataset to avoid overfitting. The preprocessing also served to improve the performance of the model by enabling the model to focus on the important features that classify stress accurately. Several machine learning algorithms were used including logistic regression, decision tree, random forest, K-Nearest Neighbors (KNN), artificial neural networks (ANN), and support vector machines (SVM). The models were compared in terms of classification accuracy. Random forest gave a maximum accuracy of 99.95% among others and was robust for stress classification while logistic regression obtained the least-precise score of 56.86%.