A new modeling method of photoplethysmography signal based on lognormal basis

Abstract

Human photoplethysmography (PPG) signal carries abundant physio-logical and pathological information of cardiovascular system, which can be used to monitor cardiovascular health in the daily life. The existing modeling methods are mainly based on Gaussian basis, which fail to conform to the long-tail features of PPG pulse waveforms. And other several existing methods based on Lognormal basis don’t work well in daily monitoring. In this paper, we proposed a new modeling method based on the long-tail Lognormal basis. Fitting calculations get an adaptive time domain by introducing the mode of the corresponding Lognormal basis and are implemented by the proposed successive-fitting solution. The simulations have proved that the proposed method has a good fitting accuracy and efficiency and is suitable for daily monitoring of cardiovascular health in body sensor networks (BSNs). Besides that, a closer relation between the cardiovascular health and the vector parameters of the Lognormal basis also can be expected.