Multivariate ARMA spectral decomposition in the assessment of cardiovascular variabilities

Multi-variate spectral analysis is able to describe the interactions between heart rate and arterial pressure variabilities; therefore, it provides a spectral decomposition based on which signal is driven more directly or on which closed-loop resonance is involved. So, it provides further insight in the genesis of rhythms, beyond the classical definition of low frequency (LF) and high frequency (HF) components related to mono-variate spectral analysis. The method of spectral decomposition is presented both for the identification of bi-variate autoregressive models, which is a general signal processing tool, and for a dynamic adjustment model specific for cardiovascular variabilities. Preliminary results on conscious dogs under various sympathetic stimuli enhancing LF rhythms confirm the existence of different mechanisms which contribute to these waves.