Modeling a stress signal

Stress is a major health problem in our world today. For this reason, it is important to gain an objective understanding of how average individuals respond to real-life events they observe in environments they encounter. Our aim is to estimate an objective stress signal for an observer of a real-world environment stimulated by meditation. A computational stress signal predictor system is proposed which was developed based on a support vector machine, genetic algorithm and an artificial neural network to predict the stress signal from a real-world data set. The data set comprised of physiological and physical sensor response signals for stress over the time of the meditation activity. A support vector machine based individual-independent classification model was developed to determine the overall shape of the stress signal and results suggested that it matched the curves formed by a linear function, a symmetric saturating linear function and a hyperbolic tangent function. Using this information of the shape of the stress signal, an artificial neural network based stress signal predictor was developed. Compared to the curves formed from a linear function, symmetric saturating linear function and hyperbolic tangent function, the stress signal produced by the stress signal predictor for the observers was the most similar to the curve formed by a hyperbolic tangent function with p < 0.01 according to statistical analysis. The research presented in this paper is a new dimension in stress research - it investigates developing an objective stress measure that is dependent on time.