差分自回归移动平均与广义回归神经网络组合模型在丙型肝炎月发病率中的预测应用

发布时间：2018-11-01 14:09

【摘要】：目的探讨差分自回归移动平均(ARIMA)与广义回归神经网络(GRNN)组合模型在丙型肝炎月发病率中预测建模效果及应用前景,为疫情预测提供依据。方法 2015年5月—2016年5月,选取山东省疾病预防控制中心法定传染病直报系统2004—2014年丙型肝炎月度发病率数据及山东省统计局发布的同期人口资料。对2004—2014年山东省丙型肝炎月发病率数据构建ARIMA模型,验证拟合精度并外推预测;将ARIMA模型拟合值作为GRNN模型的输入,实际值作为GRNN模型的输出,对样本进行训练和预测。比较单纯ARIMA模型和ARIMA-GRNN组合模型在丙型肝炎月发病率中的预测效果。结果 2004—2014年山东省丙型肝炎年均发病率为17.28/10万,并随着时间的推移呈上升趋势(Z=29.05,P0.01)。ARIMA(1,2,1)模型预测2014年山东省丙型肝炎发病率与实际发病率基本一致,落在95%置信区间内,拟合效果较好。以ARIMA(1,2,1)模型拟合值作为GRNN模型的输入,丙型肝炎月发病率实际值作为GRNN模型的输出,取最优光滑因子0.12训练模型,ARIMA-GRNN组合模型预测的拟合值与实际值基本吻合。ARIMA模型和ARIMA-GRNN组合模型的平均误差率(MER)分别为16.87%、15.30%;决定系数(R~2)分别为0.53、0.60;平均绝对误差(MAE)分别为0.17、0.09;平均绝对百分误差(MAPE)分别为1.18、0.35。结论 ARIMA-GRNN组合模型对山东省丙型肝炎月发病率拟合及预测效果优于单纯ARIMA模型,具有较高的拟合精度,有较为广阔的应用前景,对于疫情预测工作有一定的实用性意义。
[Abstract]:Objective to explore the effect and application prospect of differential autoregressive moving average (ARIMA) model combined with generalized regression neural network (GRNN) model in predicting the monthly incidence of hepatitis C. Methods from May 2015 to May 2016, the monthly incidence data of hepatitis C from 2004 to 2014 and the population data of the same period published by Shandong Bureau of Statistics were selected. Based on the monthly incidence data of hepatitis C in Shandong Province from 2004 to 2014, the ARIMA model was constructed to verify the fitting accuracy and extrapolate prediction, and the fitting value of ARIMA model was taken as the input of GRNN model and the actual value was taken as the output of GRNN model to train and predict the samples. To compare the predictive effect of simple ARIMA model and ARIMA-GRNN combination model in the monthly incidence of hepatitis C. Results the average annual incidence of hepatitis C in Shandong Province from 2004 to 2014 was 17.28 / 100 000, and showed an increasing trend with the passage of time. The results are as follows: 1) the incidence of hepatitis C in Shandong Province in 2014 is basically consistent with the actual incidence, falling within 95% confidence interval, and the fitting effect is good. The fitting value of ARIMA model was taken as the input of GRNN model, the actual value of monthly incidence of hepatitis C was taken as the output of GRNN model, and the optimal smoothing factor 0.12 training model was selected. The average error rate of ARIMA model and ARIMA-GRNN combination model is 16.8715.30, respectively. The average absolute error (MAE) is 0.170.09, and the average absolute percent error (MAPE) is 1.18 ~ 0.35. The determination coefficient (R _ (2) is 0.53 ~ (0.60), the average absolute error (MAE) is 0.17 ~ (0.09), respectively. Conclusion the ARIMA-GRNN combination model is superior to the simple ARIMA model in fitting and predicting the incidence of hepatitis C in Shandong province. It has a higher fitting precision and has a broad application prospect. It has some practical significance for the prediction of the epidemic situation.
【作者单位】：潍坊医学院公共卫生与管理学院卫生统计学教研室;
【基金】：“健康山东”重大社会风险预测与治理协同创新中心资助课题(XT-1402001)
【分类号】：R512.63

【参考文献】