Research On Heart Rate Variability Analysis Based On Poincare Plot And Recurrence Plot

Heart rate variability (HRV) analysis is the main method which is currently approved for clinical quantitative evaluation of autonomic function. As a noninvasive detection method, HRV analysis has important clinical value for the diagnosis of heart disease and other diseases associated with autonomic nervous system. HRV analysis methods usually include time domain, frequency domain and nonlinear methods, among which nonlinear methods are considered more helpful to reveal the essential characteristics of the cardiac dynamic system. However, most nonlinear methods are not mature enough and have not been widely accepted in clinical application. Therefore, to search for the nonlinear parameters which can properly reflect the differences of cardiac dynamic system under various physiological and pathological states, and then reveal the nature of the cardiac dynamic system, is the goal of many researchers working on HRV.In the nonlinear analysis of HRV, graphical phase space analysis represented by Poincare plot and recurrence plot (RP) are distinctive and effective, which can visually exhibit the nonlinear characteristics of heart rate fluctuations. However, the objectivity and accuracy of the analysis results cannot be guaranteed. Thus the problem lies in capturing information of these plots that can reflect the nonlinear dynamic characteristics of HRV time series quantitatively.Focusing on these problems, we have conducted some research and exploration on methods of heart rate variability analysis based on Poincare plot and RPs. Some novel quantitative measures to reflect the nonlinear characteristics of these plots are presented in this work. According to these measures, we studied the differences of nonlinear dynamics of HRV series in different physiological and pathological states. Some meaningful results which reflect the complexity and nonlinear characteristics of human cardiac autonomic nervous activity were obtained in our study. The specific innovative work includes the following aspects:(1) Based on the second-order difference plot, we propose a novel quantitative measure-multi-scale distribution entropy (DE), to reflect the distribution properties in annular regions of different radii and in four different quadrants separately. Furthermore, the dynamics of the plot on multiple time scales can be comprehensively studied. The method was applied to the RR interval series derived from healthy young subjects, old subjects and subjects with congestive heart failure (CHF) respectively, to investigate the impacts on HRV due to age and typical pathological states. The results show that, the differences of the DE results between groups are most significant in quadrant I, which highlights the impact on vagal regulation (dysfunction or damage) brought by aging or disease. The DE values change with the scale factor in certain regularities, indicating dynamics with long-range correlation on multiple temporal scales of the heart rate control system. Regularities for healthy subjects and CHF sufferers are significantly different, reflecting the full impact of the pathological states on the multi-scale heart rate regulation mechanism. The algorithm was further validated by simulated noises and surrogate data test. Results show that it indeed reflects some intrinsic features of the physiological systems independent of the mean and standard deviation of the series. In addition, on certain scales, the difference of DE in quadrant I (ED1) between healthy and CHF groups is most significant. Therefore, the ED1on those scales may be used as new clinical indices for the auxiliary diagnosis of heart failure. Finally, the RR interval series of healthy subjects in daytime (awake) and nighttime (asleep) were analyzed respectively, to investigate the circadian rhythm of the autonomic nervous regulation. Results show that, the differences of DE values and their dependences on time scales can effectively reflect the changes of autonomic regulation related with circadian rhythms.(2) According to characteristics of the second-order difference plot, we propose the concept of nonlinear feedback. Meanwhile, taking into account that the heart rate is regulated by various mechanisms via multiple feedback loops incorporating different delays, another new method named multi-scale feedback ratio analysis is proposed for the quantitative analysis of second-order difference plot, which can visually analyze the variation of positive feedback ratio, negative feedback ratio and the overall proportion with the increase of the temporal scale. With this method, we first analyzed three kinds of typical simulated noises-white,1/f and Brownian noises. Results show that, their feedback ratios are indeed on different levels. Taking this result as a reference, the method is then applied to three kinds of different RR interval series derived from healthy subjects, subjects with CHF and subjects with atrial fibrillation (AF) respectively. Results show that, on relatively small time scales, their variation of multi-scale feedback ratios are significantly different; on relatively large time scales, the feedback levels of the three groups are also significantly different, and close to those of the1/f, Brownian and white noises respectively, indicating different feedback mechanisms. In addition, the quantitative indicator RTF extracted from the curve is sensitive to the difference between the groups, and thus can serve as a useful supplement for the existing reference criterion of the plot. Finally, we studied the multi-scale feedback properties of synthetic RR interval series. It is found that on small scales, their feedback levels and variation with time scales are significantly different from those of physiological RR interval series, reflecting the deficiencies of existing artificial HRV models. The quantitative indicator RTF extracted for evaluation of mean feedback levels on small time scales has a certain effect for differentiating between synthetic and physiological RR interval series, and thus can be used as a tool for testing models of HRV signals.(3) As to the application of RP in HRV analysis, most of the recent researchers have been focused on the specific applications of traditional recurrence quantification analysis (RQA) methods. However, when applying for HRV analysis, the existing RQA measures are not sensitive enough for distinguishing certain physiological and pathological states. Therefore, the improved method-multi-scale RQA is proposed in this paper. With this method, the RQA measures on multiple coarse-grained scales or wavelet scales can be investigated to select the characteristic scale on which differences between healthy and other pathological states are most significant. Therefore the value of RQA measures in HRV analysis is enhanced. We applied the multi-scale RQA method on HRV series of healthy subjects and subjects with CHF. Results show that, within a specific range of scales, the recurrence characteristics of abnormal heart rate fluctuations of CHF sufferers are more prominent than those on original scale, and thus can be better reflected by RQA measures. Furthermore, we applied the multi-scale RQA method on RR interval series of healthy subjects before and during meditation. Results show that, during meditation, RQA measures and their dependencies with time scales are quite different from those before meditation. It reflects that the meditation process has effectively changed the autonomic nervous regulation states of the subjects on multiple temporal scales, and improved the sympathetic-vagal homeostasis. Therefore, the overall states of life activities are optimized during meditation. In addition, in the field of sports, there are few applications of HRV analysis with RP method. Therefore, in this research, a group of HRV data from healthy young people during rope skipping was collected and analyzed with RP and RQA. Results show that, several RQA measures can reflect the changes of cardiac sympathetic and vagal tone during exercise and recovery processes. Thus it provides some new ideas for the expansion of HRV analysis in sports fields.