关键词: 时间序列分析, 季节滑动平均混合模型法, 预测, 疟疾, 发病率

Abstract:

Objective To explore the feasibility of the application of time series model to predicting the malaria incidence in China so as to provide the theoretical basis for the reduction of malaria incidence. Methods An autoregressive integrated seasonal moving average (ARIMA) model was established based on the data on malaria incidences from 2004 to 2009 in China. With the difference method used to smooth the sequence, we determined the order and established the 2010 national malaria incidence forecast model to evaluate the predicting results. Results It was found that there was a year-by-year decrease in the incidence and death rate of malaria from 2004 to 2009, and the incidence fluctuated with seasons with the peak incidence seen from July to October. Seasonal ARIMA model fitted well, and the residual autocorrelation function graph showed that the residuals were white noise sequences, which fitted well with the incidence sequences in the previous periods of time. Further prediction of the incidences in the individual months in 2010 with the model indicated that malaria incidence decreased in the year compared with the previous years. Conclusion The seasonal ARIMA model can be well applied to the fitting of data on malaria incidence with good forecasting, providing a scientific basis for the development of measures for the prevention and control of malaria.

Key words: Time series analysis, Seasonal ARIMA model, Forecasting, Malaria, Incidence