Estimation of Sympathetic and Parasympathetic Level during Orthostatic Stress using Artificial Neural Networks

This study deals with the development of a new method to quantify the effect of orthostatic stress on the cardiovascular system. Orthostatic hypotension in healthy subjects triggers the baroreflex, which induces increased sympathetic activity and decreased parasympathetic activity. We performed a tilt-table test on 19 healthy subjects while measuring electrocardiogram, galvanic skin resistance and blood pressure signals. We developed a method for inverse parameters identification using artificial neural networks to fit the experimental data and identify physiological parameters (sympathetic and parasympathetic level). We implemented a supervised controller in the form of mathematical model of the baroreflex which was used to estimate the sympathetic and parasympathetic levels for a selected set of experimental data. Obtained result was used as training set for our artificial neural network. The network was able to estimate the levels of sympathetic and parasympathetic discharge. The estimated values were successfully validated against the measured heart rate signal with low least-square error. Additionally we proposed a classifier, which was able to predict the sex, age, weight and health status of the patient based on estimated model parameters.