弹性反向传播神经网络在冠心病发病率预报中的应用

摘要

目的 探讨弹性反向传播(BP)神经网络在时间序列资料分析中的应用,建立冠心病发病率的预测模型.方法 利用青海海西州地区2003～2009年登记的冠心病发病率时间序列资料,以双曲正切S型函数为传输函数、隐层节点为4与11的三层BP神经网络,建立两种冠心病发病率的非线性时间序列预测模型.结果 建立的ANN2预测模型简单易行,预测值平均相对误差为0.005 547,预测精度高.结论 BP人工神经网络可以用于冠心病发病率的预测.%Objective To discuss the application of resilient back propagation (BP) artificial neural network predictive model and establish predictive incidence rate model of coronary artery heart disease( CAD). Methods Two nonlinear time series model of resilient BP artificial network were established for the incidence rate of CAD from 2003 to 2009 in Haixizhou region, Qinghai province of China, using three layers BP neural network based on hyperbolic logarithm function with the number of hidden node four and eleven. Results ANN2 predictive model was simpler and more effective with high prediction precision, the average relative error value was 0. 005 547. Conclusions The resilient BP artificial neural network predictive model could be used to forecast CAD incidence rate with high prediction precision of short-term time series.