基于短时心率变异性信号两种熵测度的过速型室性心律失常的预测分析

Predictive analysis on ventricular tachyarrhythmia using two entropy measures of short-term heart rate variability signals

目的:过速型室性心律失常[持续性室性心动过速或心室纤颤(VT/VF)]是心脏猝死的主要诱因,测试VT/VF发生前心率变异性信号是否有明显改变可作为VT和VF发生的提前预报信号。方法:以78名患者体内心脏复律除颤器记录的VT/VF事件发生前心率变异性信号(VT/VF序列)和来自同一患者的正常窦性节律(CON序列)组成的135个样本对作为实验序列。通过预处理消除实验序列的伪差、异位心搏等干扰,采用两种基于熵的非线性复杂度测度——样本熵和逐点多尺度熵(PPMSE),分析VT和VF发生前十几分钟的VT/VF序列,以及心率增加和减小的VT/VF序列复杂性,并采用PPMSE方法讨论了接近VT/VF发生时VT/VF序列复杂性变化。结果:与正常对照组CON序列相比,在一定匹配容差内,VT/VF发生前心率变异性信号的样本熵明显减小(r<0.25×SD,P<0.0005),心率增加的VT/VF序列减小更显著(r<0.3×SD,P<0.0001);VT/VF序列的PPMSE在越接近VT/VF发生时刻减小越显著,提取的CI指数存在显著差异(如1~30尺度,N=986、500、250时,P=1.5×10^-2、P=4.3×10^-3、P=1.3×10^-5),心率增加的VT/VF序列区分性能更好。结论:过速型心律失常的自然发作并不是突发现象,在其发作前或许存在某种生理预兆,两种熵测度可能是短时预报恶性室性心律失常事件的有效非线性参数。

Objective The ventricular tachyarrhythmia including sustained ventricular tachycardia (VT) and ventricular fibrillation (VF) is the major precipitating factor of sudden cardiac death. Whether there is significant change in the heart rate variability (HRV) preceding the onset of VT/VF events can be tested and the results can be served as the early warning signs of VT or VF. Methods A total of 135 samples consisted of heart rate variability signals preceding the onset of VT/VF events (VT/VF series) and normal sinus rhythm (CON series) from the same patient were taken as test series, and these samples were recorded by implantable cardioverter defibrillators (ICD) in 78 patients. The interferences such as artifacts and ectopic beats in test series were removed by signal pre-processing, and two nonlinear complexity measures based on sample entropy and point by point multiscale entropy (PPMSE) were used for complexity analysis on VT/VF series with more than 10 minutes preceding the onset of VT/VF events, and VT/VF series of increased or decreased heart rates preceding the onset of VT/VF events. The complexity of VT/VF series when gradually approaching VT/VF occurrence was also discussed based on PPMSE multi-scale entropy method. Results Compared with that of CON series, in some range of matching tolerance, the sample entropy of HRV preceding the onset of VT/VF events was significantly decreased (r<0.25×SD, P<0.000 5), and the decrease was more obvious for VT or VF series of increased heart rate (r<0.3×SD, P<0.000 1). The decrease of PPMSE of VT/VF series were more significant when gradually approaching VT/VF occurrence, and there was significant difference for the extracted CI value (such as for scale range of 1-30 and N=986, 500, 250, the P value was 1.5×10^-2, 4.3×10^-3, 1.3×10^-5, respectively), and the differentiation for VT/VF series of increased heart rates was even better. Conclusion The spontaneous occurrence of ventricular tachyarrhythmia is not a sudden phenomenon, and there may be some physiological signs before the onset. The two entropy measures might be the suitable nonlinear parameters for the short-term prediction of life-threatening ventricular tachyarrhythmia.