论文部分内容阅读
使用计算机建立心电波离散系统自回归(AR)模型,可用于心电波的诊断和数据的压缩。所用的心电波包括正常和异常的(前心室收缩)二大类,每一类都有40组采样数据,每一组数据各包含4个心电波周期。将每组数据经模数转换成数字量,其采样周期为4毫秒。通过最小二乘法,求出适配于每组数据的线性的不随时间而变的自回归模型。并求出正常心电波的数学模型最佳阶次为6,异常心电波最佳阶次为5。最后使用各种统计方法来检验所获得的模型的有效性。 在验证模型的有效性时,为便于比较,二类心电波均采用6阶模型。使用了均值为零的检验法验证模型公式的残余误差,用Portmanteau检验法验证模型残余误差的白色度;求出残余误差的相关系数验证残余误差是否代表一个白色噪音过程。检验结果表明,6阶模型能满意通过均值为零和相关系数的检验,而对于Portmanteau检验,只有当模型阶次提高到19阶时,大部分的模型才能通过检验。因此这个检验方法在这种特定应用中显得非常严格。为了验证模型的有效性,又将从测量数据求出的实验相关图和从6阶模型的输出数据求得的“理论”相关图进行比较。结果表明,二种相关图能很好适配。 由于6阶模型参数的数值,对于正常和异常二大类心电波是明显不同的,因此这种模型有可能用于帮助心电波的诊断,并可?
The use of computer to establish the system of cardiac arrhythmia discrete autoregression (AR) can be used for the diagnosis of ECG and data compression. The heartbeats used include two broad categories of normal and abnormal (pre-ventricular contractions), each of which has 40 sets of sample data, each containing 4 cycles of ECG. Each group of data through the analog to digital conversion, the sampling period of 4 milliseconds. Using the least-squares method, a linear, time-independent autoregressive model adapted to each set of data was obtained. The best order of the mathematical model of normal ECG is 6, and the best order of abnormal ECG is 5. Finally, various statistical methods were used to test the validity of the model obtained. To verify the validity of the model, for the purpose of comparison, the 6th order model is used for both types of ECG. The residual error of the model formula was verified by the test with mean zero. Portmanteau test was used to verify the whiteness of the model residual error. The correlation coefficient of the residual error was calculated to verify whether the residual error represented a white noise process. The test results show that the 6th order model can satisfactorily pass the test with zero mean and correlation coefficient, while for the Portmanteau test, most of the models can pass the test only when the model order is increased to 19th order. So this test method in this particular application is very strict. In order to verify the validity of the model, the experimental correlation graph obtained from the measurement data is compared with the “theoretical” correlation graph obtained from the output data of the sixth-order model. The results show that the two correlation diagrams fit well. Since the values of the 6th order model parameters are significantly different for the two categories of normal and abnormal ECGs, this model may be used to aid in the diagnosis of ECG waves and can?