论文部分内容阅读
表征极短时(<1min)心率变异性(HRV)的分析参数通常随时间呈现出复杂的变化模式,这种随时间变化的波动很容易干扰对心血管状态的正确判断。本文将年龄匹配的41例健康人(对照组)和25例充血性心力衰竭(CHF)患者(实验组)的长时HRV序列划分成多段极短时序列,计算同一HRV参数在多段极短时序列中的变异系数以及多次组间t检验中差异有统计学意义的检验比例,以此探讨部分极短时HRV分析参数在揭示不同状态下心血管系统功能差异时的稳定性;此外,通过对受试者工作特征(ROC)曲线的分析以及人工神经网络的建模,评估了这些参数对对照组和实验组进行分类的效果。本文结果表明:1基于复杂网络分析的度分布熵指标有着最小的变异系数且对病理状态敏感(79.75%情况下对照组和实验组的差异有统计学意义),可为临床医生提供一个诊断CHF患者的辅助指标;2将庞加莱散点图进行椭圆拟合后,对照组和实验组的椭圆短长轴之比(SDratio)在98.5%的情况下差异有统计学意义;在人工神经网络建模时,仅使用SDratio对对照组和实验组进行分类的正确率为71.87%,表明SDratio或可作为CHF患者的智能诊断指标;3仍需寻找可用于极短时HRV分析研究且对CHF患者更加敏感特异的稳定指标。
Analytical parameters that characterize heart rate variability (HRV) in extremely short durations (<1 min) often exhibit complex patterns of change over time, and this fluctuation over time can easily disrupt the correct assessment of cardiovascular status. In this paper, long-term HRV sequences of 41 age-matched healthy subjects (control group) and 25 patients with congestive heart failure (experimental group) were divided into multistep extremely short time series, and the same HRV parameters The coefficients of variation in the sequences and the test scores in the t-test between groups were statistically significant, so as to explore the stability of some very short HRV analysis parameters in revealing the differences in cardiovascular system function in different states. In addition, The analysis of ROC curves and the modeling of artificial neural networks evaluated the effect of these parameters on the classification of control and experimental groups. The results of this paper are as follows: 1 The distribution of entropy index based on complex network analysis has the smallest coefficient of variation and is sensitive to the pathological state (79.75%, the difference between the control group and the experimental group is statistically significant), providing clinicians with a diagnosis of CHF (2) After Poincaré scatter plots were fitted by ellipsometry, the ratio of long axis of ellipse to short axis of control group was significantly different from that of experimental group (98.5%); in artificial neural network SDratio was used to classify the control and experimental groups using only 71.87% of the SDratio, indicating that SDratio can be used as an intelligent diagnostic marker for CHF; 3 there is still a need to find an instrument that can be used for very short HRV analyzes and for patients with CHF More sensitive and specific stability indicators.