Application of a Hybrid Model for Predicting the Incidence of Tuberculosis in Hubei, China

摘要: Background A prediction model for tuberculosis incidence is needed in China which may be used as a decision-supportive tool planning health interventions and allocating resources. Methods The autoregressive integrated moving average (ARIMA) was first constructed with the data of report rate Hubei Province from Jan 2004 to Dec 2011.The 2012 Jun were validate model. Then generalized regression neural network (GRNN)-ARIMA combination established based on ARIMA Finally, fitting accuracy two models evaluated. Results total 465,960 cases reported between 2011 Province. The highest 2005 (119.932 per 100,000 population) lowest 2010 (84.724 population). time series show gradual secular decline striking seasonal variation. (2, 1, 0) × (0, 1)12 selected several plausible models. residual mean square error GRNN-ARIMA 0.4467 0.6521 training part, 0.0958 0.1133 validation respectively. absolute percentage hybrid also less than model. Discussion Conclusions attributed effect intensive measures tuberculosis. variation have resulted factors. We suppose that delay surveillance system contributed According accuracy, outperforms traditional model, facilitate allocation resources China.