Application of time series analysis in the prediction of incidence trend of influenza-like illness in Shanghai

Abstract

To forecast incidence trend of influenza-like illness in Shanghai. We collected everyday-report influenza-like illness surveillance information from January, 2004 to April, 2006 and used autoregressive integrated moving average model (ARIMA) to analyze and establish prediction model. 114 weeks preceding information was used to establish model and 9 weeks data to evaluate. Model ARIMA (1,0,0) (1,1,0) 26 from Surveillance information was both with seasonal and non-seasonal features (P < 0.001). White noise analysis show the minimum Box-Ljung value of autocorrelation function was 0.803 (P > 0.1) and the residual was randomized difference. We established prediction model as lgY(t) = 0.879 lgY(t-1) + 0.418 lgY(t-26) - 0.367 lgY(t-27) + 0.582 lgY(t-52) - 0.512 lgY(t-53) and forecasting effect was well. True values were all between 95% CI of predicted ones. ARIMA model can be well used to simulate incidence trend of influenza-like illness in Shanghai.