将ARIMA模型应用于深圳市1980-2007年重点法定传染病预测分析

目的]探讨自回归移动平均混合模型（ARIMA模型）在预测法定传染病发病方面的可行性，为传染病预防控制提供科学依据。方法]选取深圳市发病数高、疾病负担较重、有代表性的部分重点传染病，利用时间序列的ARIMA模型对1980～2007年的发病数进行拟合。结果]ARIMA模型对病毒性肝炎、狂犬病、细菌性痢疾、肺结核、伤寒与副伤寒、淋病、梅毒、猩红热、艾滋病病毒携带与艾滋病、风疹的拟合效果较好，对霍乱、疟疾、流感、流脑、腮腺炎、麻疹、登革热的拟合效果较差。结论]ARIMA模型对深圳市几种重点传染病发病的时间序列变动趋势能进行较好的模拟．因此可以为法定传染病的预测提供依据。

Feasibility Study on ARIMA Model in the Prediction of the Key Notifiable Communicable Diseases in Shenzhen City from 1980 to 2007

Objective]To explore the feasibility of ARIMA model in the prediction of the prevalence of key notifiable communicable diseases and to provide scientific basis for the control and prevention of communicable diseases. Methods] High prevalence of communicable diseases, which resulted in heavy burden and the representative diseases were selected. The number of the cases from 1980 to 2007 were fitted with time series prediction method of ARIMA. Results]The fitting effect with ARIMA model was good in the prediction of several notifiabie diseases including viral hepatitis, human rabies,dysentery, tuberculosis, typhus fever and paratyphoid fever, gonorrhea, syphilis, scarlet fever, HIV infection and AIDS and rubella, however the fitting effect with ARIMA was not good for the prediction of cholera, malaria,influenza, epidemic meningitis,mumps.measles and dengue fever. Conclusion]The change of time series of the prevalence of norifiablc communicable diseases can be simulated with ARIMA model, which can provide evidence for the prediction of notifiable communicable diseases.