Assessment of Stress Level with Help of “Smart Clothing” Sensors, Heart Rate Variability-Based Markers and Machine Learning Algorithms †

Abstract

Physiological stress in healthy subjects was assessed using heart rate (HR), monitored with the help of Hexoskin smart garments. HRs were collected during daily life activities and in laboratory settings during stress tests. Heart rate variability parameters were computed and referenced with expert levels of stress. The data were processed with the help of machine learning algorithms (Random Forest, CatBoost, XGB, LGBM, SVR). The Random Forest Regressor provided the best rate of correct entries (86%), and the CatBoost Regressor provided the shortest time (2 ms) for the assessment of stress levels.