Objective To investigate the application of the seasonal differential autoregressive moving average(SARIMA)model and back propagation neural network(BPNN)model in the incidence of tuberculosis in Jiangsu Province and to compare and evaluate the accuracy of the two models to provide a reference for the development of tuberculosis prevention and control strategies in Jiangsu Province.Methods The SARIMA and BPNN models were established by using the tuberculosis incidence data in Jiangsu Province from 2011 to 2020.The actual tuberculosis incidence from January 2021 to June 2022 was tested,and the prediction accuracy and modeling effect of the two models were compared.Results The root mean squared error(RMSE),mean absolute error(MAE),mean absolute percentage error(MAPE),overall relative error value,and minimum error rate of the SARIMA model were 192,128,5.18％,0.05％,and 1.14％.Moreover,the RMSE,MAE,MAPE,overall relative error value,and the minimum error rate of the BPNN model were 301,246,11.03％,2.79％,and 1.44％.Conclusions The SARIMA model can fit and predict the monthly incidence rate of pulmonary tuberculosis in Jiangsu Province quite well,providing a basis for monitoring and controlling the disease.
NETL
NSTL
万方数据
