The analysis of heart sounds based on linear and high order statistical methods

Abstract

This paper investigates the applicability of high order statistical autoregressive (AR-HOS) modeling method in analyzing biomedical signals. The autoregressive (AR) method using linear prediction and AR-HOS method using cumulants are applied on normal and pathological heart sound signals. It is found that the AR-HOS modeling a signal produce more accurate and higher resolution spectrum than AR modeling.