Exercise evaluation from Blood Volume Pulse signals analyzed by parametric auto-regressive modeling

Abstract

It has been reported in the past that the Blood Volume Pulse (BVP) waveform, and in particular its dicrotic notch, reflects changes of the cardiovascular system when a person is subjected to physical stress. The dicrotic notch becomes less prominent with exercise, disappearing completely in some cases and reappears after the subject rests for some time. Quantification of the changes in features of the BVP waveform, such as the dicrotic notch, could yield a single numerical parameter to indicate the amount of exercise a subject has undergone. This paper reports the automation of an algorithm using autoregressive modeling to get such numerical index. For this study, BVP measurements from 15 subjects were taken at three stages: before exercise, just after exercise and after rest. The automated algorithm took these signals as input and resulted in a numerical index for each stage. The experimental results show that autoregressive (AR) modeling and Power Spectral Density (PSD) analysis can be used to automatically obtain a number that reflects changes in the cardiovascular system of an individual with exercise. © 2009 Springer Berlin Heidelberg.