Abstract： Objective To establish a seasonal autoregressive integrated moving average (SARIMA) model for forecasting influenza in Hunan Province, and to provide an important reference basis for timely taking targeted prevention and control measures. Methods The data about influenza-like illness proportion (ILI%) in Hunan Province from the 1st week of 2010 to the 52nd week of 2015 were collected for the construction of SARIMA model. The established model was verified based on the ILI% data from the 1st week to the 16th week in 2016, and then the ILI% from the 17th week to the 52nd week in 2016 was forecasted with the model established. The module of forecasting in SPSS 18.0 software was used in the analysis. Results SARIMA (3,0,3) (1,1,0)52 was established for prediction of the trend of ILI% in Hunan Province. The stable model parameter of the Normalized BIC value was -1.048, and the residuals of the model were white noise (Ljung-Box Q=17.980, P>0.05). The absolute error between predicted and actual values ranged from 0.14% to 1.60%, and the mean absolute percentage error (MAPE) was 15.96%. Based on the predication of the model, the ILI% in Hunan Province in 2016 showed two peaks, the first peak of the epidemic would be around in the 3rd week to the 7th week and the second peak around in the 22nd week to the 27th week, which displayed the similar epidemiological characteristics with those in 2010-2015. Conclusions The established SARIMA (3,0,3) (1,1,0)52 model is of good fitting effect and can be applied in short-term forecast of weekly ILI% in Hunan Province.
周美兰, 周志华, 罗美玲, 胡世雄, 邓志红, 李桀, 王勇. 湖南省哨点医院流感样病例SARIMA模型预测[J]. 实用预防医学, 2018, 25(3): 370-373.
ZHOU Mei-lan, ZHOU Zhi-hua, LUO Mei-ling, HU Shi-xiong, DENG Zhi-hong, LI Jie, WANG Yong. Prediction of influenza-like illness in sentinel hospitalsin Hunan Province by SARIMA model. , 2018, 25(3): 370-373.
2018, Vol. 25 
