论文部分内容阅读
本文以心音特征为基础,实现了连续的左心室收缩压预测。通过对3只比格犬进行实验,以肾上腺素诱发心脏血流动力学发生变化,然后同步采集实验犬的心音、心电、左心室血压等信号,共获取了28组有效数据。通过提取心音特征,借助人工神经网络实现了反推左心室收缩血压,获得了较好的预测效果。本研究在较大的血压动态变化范围内,得到了绝对误差均值仅为7.3 mm Hg、预测血压与测量血压的平均相关系数为0.92的实验结果。研究结果显示,本文所述方法有助于实现无创的左心室血流动力的连续监测。
In this paper, based on the characteristics of heart sound, a continuous prediction of left ventricular systolic pressure was achieved. Through experiments on three Beagle dogs, epinephrine induced changes of cardiac hemodynamics, and then simultaneously collected heart sound, ECG, left ventricular blood pressure and other signals of experimental dogs, a total of 28 sets of valid data were obtained. By extracting the characteristics of heart sounds, the systolic pressure of left ventricular systolic blood pressure was achieved by means of artificial neural network, and a better prediction effect was obtained. In this study, we obtained the experimental results that the mean absolute error is only 7.3 mm Hg and the average correlation coefficient between predicted blood pressure and measured blood pressure is 0.92 within the range of larger blood pressure dynamic changes. The results show that the method described in this article helps to achieve non-invasive continuous monitoring of left ventricular hemodynamics.