Graphical Analyses Of Cardiovascular Time Series And Application

The morbidity of cardiovascular disease(CVD)is rising rapidly especially in the young adults,leading to enormous healthcare burden to both families and the society.It is believed that interventions at early or preclinical stages of CVD can better help reduce its incidence or slow the progression of the disease.Early detection techniques that are both accurate and non-invasive are thus required.To achieve this,many endeavors have been made in the past two or three decades.The graphical analysis of cardiovascular time series,such as the Poincare plot analysis,shows up as a promising tool.Poincare plot can be used to characterize the internal structure of physiological time series and describe the geometric construction of the phase space of the dynamic system.It has been widely used in the anlaysis of cardiac dynamics.However,the existing quantitative analysis methods using the Poincare plot limits to a certain aspect of the mathematical or geometrical features of the graph.Besides,none of the existing approaches is designed for multidimensional signals.In this dissertation,several new quantitative graphical analysis algorithms of the Poincare plot were developed.Influences of different factors on these algorithms were studied.Besides,a multimodal graphical analyzing framework was also established.The purpose of this dissertation was to investigate the value of these graphfical tools of short-term cardiovascular time series in the early detection of CVDs.The main work and innovations are as follows.(1)A novel area index to measure the heart rate asymmetry was developed.Traditional metrics almost exclusively focus only on one dimension,such as the number of points,distances of points to the diagonal line,namely the line of identity,or phase angles of points to the line of identity.An area index was proposed to combine information from both the distance and the phase angle dimensions.Experimental results of both long-and short-term time series showed that the area index was more robust than existing measures in terms of the ability to tell different groups apart(e.g.,the p value,area under the receiver-operating curve,and effect size).The heart rate asymmetry was increased in patients with congestive heart failure and arrhythmia,suggesting potentially a disease induced dysregulation of the cardiovascular autonomic control.Four influence points and two reference lines were proposed to studied the influencial factors on analyzing heart rate asymmetry and results of heart rate asymmetry analysis were thereby optimized.Different reference points/lines significantly affect the results of the area index and the slope index and thus should be taken into consideration.(2)The regularity of heart rate variability in patients with coronary heart disease after percutaneous coronary intervention was explored.The area index of patients with coronary artery disease was significantly decreased within 24 hours after percutaneous coronary intervention surgery than that before the surgery,suggesting that a degree of recovery of cardiovascular status in patients on the first day after surgery.It may be that the area index could obtain more information hidden in the short-term cardiac cycle time series by combining two dimensional features.The area index is potentially capable of monitoring the functional status of the cardiovascular autonomic control after percutaneous coronary intervention.The results of linear mixed model analysis indicated that the change of all heart rate variability metrics was independent of sex.(3)A novel gridded Poincare plot was proposed and quantitative algorithms to analyze the gridded Poincare plot,namely grid distribution rate and grid distribution entropy,were developed.The grid distribution rate was proposed to evaluate the scatter distribution characteristics of a time series.The grid distribution entropy was developed by applying the Shannon entropy algorithm and was used to quantifys the weighted average probability of scatter distribution.Simulation results showed that the two gridded Poincare plot-based algorithms had clearly improved consistency and stability in comparison with the traditional algorithms.Analysis of human data indicated that the grid distribution rate and grid distribution entropy of the healthy youth group were significantly larger than those of the healthy elderly group,suggesting a more dispersed distribution of the Poincare plot in young people which,potentially,indicates that the autonomic function of the elderly is degraded.The grid distribution rate and grid distribution entropy of the coronary artery disease group were significantly lower than those of the healthy control group,revealing that the coronary artery disease may have led to lower irregularity or complexity of the heart rate dynamics.(4)A multimodal physiological signal plot was constructed based on multichannel synchronously recorded cardiovascular time series.The multimodal physiological signal plot was constructed with the time series extracted from the pulse signal,heart sound and electrocardiogram signals.Using similar strategies as applied in the analysis of one-dimensional Poincare plot,an algorithm was proposed to measure the dispersion degree of scatter points in the multimodal physiological signal plot,that is,the distribution of cardiovascular time series from the perspective of a graphical spatial structure.The results indicated that coronary stenosis led to an increase in the spatial distribution of cardiac electro-mechanical and cardiac-vessel time series,revealing a reduction of the coupling across different cardiovascular intervals within the same cardiac cycle.Three-dimensional physiological signal analysis documented that the systolic period interval variability was relatively independent and stable,while the coupling between the diastolic period interval and the cardiac cycle was more obvious.Coronary artery disease led to the degradation of the coupling between the diastolic period interval and the cardiac cycle time series.The three-dimensional physiological signal plot is thus shown to be more accurate to describe the interaction among different cardiovascular time series.The graphical analysis of the same cardiac cycle potentially provides additional valuable information for explaining physiological or pathological changes in the cardiovascular system.