Comparison between ARMA model and LSTM deep neural network in predictive effect on onset trend of pulmonary tuberculosis in Jiashi County of Xinjiang
Objective To use the auto-regressive moving average(ARMA)model and long short term memory(LSTM)depth neural network to predict the incidence trend of pulmonary tuberculosis in Jiashi County.Methods The legal infectious disease report data in this area from January 2014 to June 2023 were collected to construct the data set,in which the onset data of pulmonary tuberculosis from January 2014 to De-cember 2021 were used to the model construction and the data from January 2022 to June 2023 were used to the model verification.The Eviews7.2 and MATLAB2023a softwares were used to construct the ARMA mode and LSTM neural network.The monthly onset number of pulmonary tuberculosis from 2022 to 2023 was pre-dicted.Results The root-mean-square error(RMSE)of the optimal ARMA model and LSTM neural network verification from January 2014 to June 2023 was 26.494 and 12.713 respectively,suggesting that the fitting effect of LSTM neural network was better than that of ARMA model.The predictive results by adopting the LSTM neural network was basically consistent with the actual onset situation.Conclusion The LSTM neural network has good fitting and predicting effect for the onset trend in Jiashi County,which could provide the theoretical reference for predicting the onset number of pulmonary tuberculosis in the future in this area.
万方数据
维普
