Signal Component Analysis by Use of a Diffusion Model and its Application for Autonomic Nerve Function Evaluation

Abstract

A novel signal component separating method by using a diffusion model has been developed. The diffusion process is conducted in a virtual time scale. Separated components are guaranteed to be orthogonal to the diffused signal. We apply this signal component method for humans' autonomic-nerve-function evaluation i.e. sympathetic/parasympathetic tension in daily life. The results show that this method is quit effective.