Multivariate autoregressive modeling of autonomic cardiovascular control in neonatal lamb.

The neonatal cardiovascular control system is a complicated interactive system which is under vigorous development at birth. From the measurement point of view the cardiovascular control is a closed-loop system. However, it can be examined on a beat-by-beat basis by analyzing circulatory-controlled variables with advanced signal analysis techniques. This paper proposes to use a multivariate autoregressive modeling technique in the analysis of several simultaneous physiological signals in order to examine interactions and inherent properties in the system. With the proposed multivariate autoregressive modeling technique, a signal is modeled as a linear combination of its own past and the past values of the other simultaneous signals plus a predictive error term of the model. The interactions in the system after the model identification are analyzed in frequency domain utilizing power spectrum estimates of the signals and signal contributions. The applicability of the proposed method was examined by a three-variable model between heart rate, blood pressure and respiration in the study of autonomic cardiovascular control in a chronic neonatal lamb model, in which the cardiovascular status was changed by using a beta-adrenergic autonomic nervous blockade. The study showed that the multivariate autoregressive modeling technique is a feasible technique in studying complicated interactions within the cardiovascular control system.