Application of multiple seasonal ARIMA model in prediction of tuberculosis incidence

Objective To explore the feasibility of multiple seasonal autoregressive integrated moving average(ARIMA) model to predict tuberculosis incidence.Methods Multiple seasonal ARIMA(p,d,q)?(P,D,Q)s model was built using tuberculosis surveillance data from January 1,2004 to June 30,2011 in Henan province,and the predictive performance was conducted and assessed using the data from July 1 to December 31,2011.Results The seasonal effect in the incidence of tuberculosis was observed from January 1,2004 to December 31,2011 in the province,and the incidence was slightly decreased over time.Multiple seasonal ARIMA(1,1,0)?(1,1,0)₁₂ model could better fit the incidence of tuberculosis over the period,and the forecast values were consistent with the actual number,with the average absolute error and the average absolute error rate of 0.317 and 4.77%,respectively.Conclusion Multiple seasonal ARIMA model could successfully fit and predict the incidence of tuberculosis,which could be applied for the prevention and control of tuberculosis.