The algebraic-combinatorial notion of recursive matrices can be used to represent and handle the basic properties of filter theory, such as perfect reconstruction and alias cancellation. On the other hand, the algebraic reinterpretation of the wavelet packet analysis by means of recursive matrices enables the alias-free detection of transients in non-stationary signal. In this paper, the authors consider an application for the study of typical non-stationary phenomena such as electrocardiograms or heart rate variability signals.