摘要
目的建立乙肝发病率(1/10万)的ARIMA-GM组合模型,并将其应用于乙肝发病率的预测,为及早发现疾病发展趋势和及时采取控制对策提供科学依据。方法收集河北省迁安市2004年1月-2012年12月的乙肝月发病率资料,应用SPSS 13.0软件进行ARIMA建模拟合;采用GM(1,1)模型对上述获得的带阈值的残差序列进行修正并构造出组合预测模型,利用此模型对该市2013年乙肝逐月发病率进行预测。结果 ARIMA(0,1,1)(0,1,1)12模型较好地拟合了乙肝的发病情况,模型的所有参数都通过统计学检验;采用阈值为3的GM(1,1)模型对残差序列进行修正,预测模型通过了精度检验(C=0.673,P=0.877),模型拟合精度为基本合格,ARIMA-GM组合模型的MAE、MAPE值均比单个模型小。利用组合模型对2013年乙肝发病率进行预测的结果显示,总体趋势与之前一致。结论 ARIMA-GM组合模型能较好地拟合乙肝发病情况,预测精度高于ARIMA季节乘积模型,且其预测结果能够为乙肝早期预测预警模型的建立提供借鉴,从而有针对性地采取相应的控制措施。
Objective To establish a combination model of autoregressive integrated moving average model and the grey dynamics （ARIMA-GM） of hepatitis B incidence rate （1/100 000） to predict the trend of outbreak of hepatitis B, as to provide a scientific basis for the early discovery of the infectious diseases for the performance of countermeasures of controlling its spread. Methods The monthly incidence of hepatitis B in Qjan＇an city, Hebei province, was collected from Jan 2004 to Dec 2012, and a model （ARIMA） was reproduced with SPSS software. The GM （1,1） model was used to correct the residual sequence with a threshold value, and a combined forecasting model was reproduced. This combination model was used to predict the monthly incidence rate in this city in 2013. Results The model ARIMA（0,1,1）（0,1,1）I2 was established successfully and the residual sequence was a white noise sequence. Then the GM （1,1） model with a threshold of 3 was used to correct its residuals and obtain its nonlinear feature extraction of information. The forecasting model met required precision standards （C=0.673, P=0.877）, the fitting accuracy of which was basically qualified. The results showed that the MAE, MAPE of the ARIMA-GM combined model were smaller than that of a single model, and the combined model could improve the prediction accuracy. Using the combined model to forecast the incidence of hepatitis B during Jan 2013 to Dec 2013, the overall trend was relatively consistent with the condition of previous years. Conclusion The ARIMA-GM combined model can better fit the incidence rate of hepatitis B with a greater accuracy than the seasonal ARIMA model. The prediction results can provide the reference for the early warnin^z system of HBV.
出处
《解放军医学杂志》
CAS
CSCD
北大核心
2014年第1期52-56,共5页
Medical Journal of Chinese People's Liberation Army
基金
河北联合大学青年科学研究基金资助项目(z201225)~~
关键词
模型
统计学
预测
肝炎
乙型
慢性
models, statistical
forecasting
hepatitis B, chronic
