Application of ARIMA model and BP neural network model in prediction of tuberculosis incidence in Gansu Province / 中华疾病控制杂志

ABSTRACT

Objective To investigate the predictive effect of autoregressive integrated moving average (ARIMA) model and back propagation neural network (BPNN)in the prediction of tuberculosis incidence in Gansu Province, and to select appropriate models to predict the incidence. Methods Based on the data of tuberculosis in Gansu Province from 1997 to 2017, the ARIMA time series model and BP neural network model were established to predict the incidence from 2018 to 2019, and the prediction accuracy and modeling effect of the two models were compared. Results For the incidence of tuberculosis in Gansu Province in 2018 and 2019, the ARIMA model predicted results were 55.1075, 54.5373, MSE=92.24, MAE=7.5313, MAPE=9.26%; BP neural network model predicted results were 62.0132, 73.4460, MSE= 9.6575, MAE = 1.1449, MAPE = 1.68%. Conclusions The BP neural network model has a better predictive effect on the incidence of tuberculosis in Gansu Province, and it shows that the incidence of tuberculosis in Gansu Province will increase slightly from 2018 to 2019.