Regression Modelling for Stress Detection in Humans by assessing most prominent Thermal Signature

摘要

Many previous studies have demonstrated non invasive and non contact methods for stress detection. However, very few studies are reported to identify stress states at individual level or regression modelling for identifying prominent features for stress detection. We, here, report a multinomial logistic regression model for stress detection based on Heart Rate Variability (HRV), for detecting stress in students, by assessing most prominent thermal signature. It models various thermal features and predicts the stress level. Thermal signatures are used to establish stress markers such as Heart Rate, HRV, so as to detect the most prominent signature for stress detection in students during competitive examination. These parameters are extracted from 10 participants and further used to classify psychophysiological state i.e. stress state, of students. Age, gender, CGP A of the students are other factors considered in this work. Likelihood ratio test is performed to find the significance of each predictor variable. Odds ratio test is made to find most significant thermal signature. Prediction classification rate is also made to evaluate the regression model. Our work shows high confidence w.r.t ground truth values.