Abstract: Objective To explore the prospect of SARIMA-RBF combination model in fitting and predicting the epidemic tendency of mumps in China. Methods SARIMA model was established using the monthly incidence data of mumps from January 2005 to December 2015 in China. The fitted value based on the SARIMA model was taken as the input vector and the actual value was used as the output vector. Two SARIMA-RBF combination models were established according to the time factor as the input vector or not. The time factor was added as the combination model A, and the time factor was not added as the combination model B. Using SARIMA model and two SARIMA-RBF combination models, the incidence rates of mumps were predicted from July to December 2016, and compared with actual values. Three model fitting and prediction effects were evaluated by using mean absolute percentage error （MAPE）, mean error rate （MER）, mean square error （MSE）, and mean absolute error （MAE）. Results The SARIMA （0, 0, 1） （0, 1, 1）₁₂ was the optimal SARIMA model. Fitted by SARIMA model, combination model A and combination model B, MAPEs were 15. 724%, 12. 217%, and 13. 941%, the MERs were 15. 168%, 10. 179%, and 14. 042%, the MSEs were 0. 336, 0. 167, and 0. 713, and the MAEs were 0. 296, 0. 199, and 0. 274,respectively; the predicted MAPEs were 12. 069%, 7. 904%, and 9. 598%, the MERs were 12. 331%, 7. 872%, and 10. 636%, the MSEs were 0. 022, 0. 013, and 0. 025, and the MAEs were 0. 138, 0. 088, and 0. 119, respectively. Conclusion The SARIMA-RBF combination model considering time factor was the optimal fit and prediction model, with application value.

Key words: SARIMA, RBF neural network, Seasonal index, Mumps