Técnicas de sistemas dinâmicos não-lineares na análise da modulação autonômica da variabilidade da freqüência cardíaca

Abstract

The heart rate variability (HRV) is a cardiovascular variable that has already established itself as a provider of relevant information on the mechanisms of autonomic control of heart activity. The complexity of the HRV signal suggests that nonlinear phenomena may play an important role in autonomic modulation of cardiac pacemaker. In fact, the literature has shown promising results in the study of pathophysiological processes associated with HRV indices when derived from nonlinear dynamical systems and deterministic chaos are used in the analysis of this signal. However, interpretation of these results is more difficult than those obtained by classical methods derived from linear systems. Aiming to better characterize the dynamics of the HRV signal, the first objective was to evaluate the presence of determinism, without assuming the presence of chaotic dynamics in the system. Confirmed the presence of determinism, it is clear the ability to use tools that originate in the area of nonlinear dynamical systems to study the mechanisms of control of the cardiovascular system. This study was conducted with a series of HRV obtained from rats previously submitted to pharmacological autonomic blockade and adult volunteers without clinical evidence of cardiovascular disease. The investigation of determinism was carried out using a new method proposed in this work, which is based on modeling and prediction NARMA free. From the series of HRV, stationary sections of short duration (64s for 5 min for rats and humans) were selected and a collection of surrogates has been generated. It is assumed that the surrogates are by construction random time series, and his prediction should not be higher than the trivial predictor, ie the average. HRV signals and their surrogates underwent NARMA modeling and prediction. The results show that the prediction of surrogates was not trivial to predict higher than average, while most of the series of HRV had a higher predictive performance (86% to 63% for rats and humans), suggesting that the series of HRV have some degree of determinism in its dynamics. It was also noted that among the series of HRV obtained from mice submitted to pharmacological autonomic blockade, fewer showed determinism (25% in series with predominant parasympathetic blockade and 40% with predominant sympathetic blockade.) Although the physiological interpretation of these results is not obvious, it could be speculated that vagal tone plays a major role on the deterministic component inherent control mechanisms of cardiac activity. This study also sought to question responded poorly investigated in the literature: the techniques of pre-processing, used to generate series of HRV with uniform sampling intervals, significantly alter the determinism and the main indices of HRV derived from dynamical systems and nonlinear Linear (paired t test, p <0.05). Two methods of preprocessing widely used in literature were used in this paper: the method based on convolution of the series of HRV with a rectangular window and the cubic polynomial interpolation method could be used in more recent studies involving HRV nonlinear techniques of analysis . After the investigation of the characteristics of the HRV signal described above, the third objective was to decompose the series into components of HRV possibly simpler, they were associated with some physiological process. Thus, the level of determinism and some HRV indexes could be calculated from these components to see if the results are more sensitive index (p less significance) to changes in cardiac autonomic activity level. The wavelet transform based on Daubechies orthogonal wavelet was used in the decomposition of HRV signals on scales that were associated with frequency bands commonly referred to mechanisms of autonomic regulation of cardiac pacemaker. The results showed that some HRV indices calculated from certain wavelet components were more sensitive to changes in autonomic tone, suggesting that these components can be useful in the characterization of autonomic modulation. The degree of determinism in this component was consistent with the results of the entire series. No pattern of distribution of relevant deterministic components was observed so that he could relate determinism with any physiological mechanisms of regulation of heart rate observed in different wavelet components. A preliminary assessment of the clinical applicability of the determinism test proposed in this work was performed by analyzing a series of HRV obtained from a patient who had sudden cardiac death during Holter examination achievement. This series was divided into segments of 5 min supposedly stationary, which were submitted to analysis of determinism. It was shown that the degree of determinism decreased as time progressed to the fatal event, while the degree of determinism in a series of control obtained from normal volunteers remained constant on average. This result suggests that the method of investigating determinism has potential in clinical medicine.