Prediction Of AIDS Incidence In China Based On ARIMA-BP Model

Objectives(1)Through analyzing the national AIDS incidence data from 2008 to 2018,to understand the epidemic trend and characteristics of AIDS incidence in the country in the past 10 years.(2)To explore the effect of the ARIMA-BP neural network combination model in the prediction of actual AIDS incidence,and combine it with the actual situation in China to establish an appropriate AIDS incidence prediction model to control the development of AIDS,formulate prevention and control measures for the country,and Allocating medical resources provides a theoretical basis.Methods(1)Establish a linear model based on the total population of the country from 2008 to 2018,and estimate the monthly population in each year.Based on the number of AIDS cases in China from 2008 to 2018,calculate the national AIDS incidence rate from 2008 to 2018.(2)The ARIMA and BP neural network models are used separately,and the incidence data from January 2008 to December 2017 are used as model fitting values to predict the monthly AIDS incidence data from January to December 2018.The value is compared with the actual value,and the accuracy of the model is measured by the size of the relative error.(3)The equal weight method,the advantage matrix method,and the tandem method are used to model and predict the combination of the two models.The error mean square,average absolute error,and average absolute percentage error are used as evaluation indicators to evaluate the prediction accuracy of all models.Results(1)From 2008 to 2009,the number of patients was relatively small,and the growth rate of the number of patients was slow;from 2009 to 2010,the number of patients increased abruptly;after 2011,the number of patients was relatively stable,showing periodic changes.(2)The MSE of single ARIMA model is 0.0021,the MAE is 0.0356,and the MAPE is 10.0636%.(3)The MSE of single BP neural network model is 0.0030,the MAE is 0.0427,and the MAPE is 11.1487%.(4)The MSE of combined model with equal weight method is 0.0024,the MAE is 0.0376,and the MAPE is 10.1644%.(5)The MSE of the advantage matrix method is 0.0023,the MAE is 0.0368,and the MAPE is 10.0996%.(6)The MSE of Tandem model combined model is 0.0016,the MAE is 0.0334,and MAPE is 9.6914%.Conclusions(1)The monthly incidence of AIDS in China during the 10-year period from 2008 to 2018 has been cyclical and has stabilized since 2010.(2)The prediction effect of a single ARIMA prediction model is excellent inshortterm prediction,which indicates that the national AIDS monthly incidence rate data has good linear characteristics and can be used as a short-term prediction model application.(3)Among the three combination models in this study,only the series method is better than the single ARIMA model.The first mock exam is that all the combined models are better than the single model.In the actual research,we should study and analyze more and choose the best combination model.(4)It is generally considered that the average absolute percentage error is less than 10%,the prediction accuracy of the model is higher,and the average absolute percentage error is less than 5%,the prediction accuracy of the model is very high.The average absolute percentage error of series method is 9.69%,which is lower than 10%.It can be considered that the prediction accuracy is high.Moreover,in the long term prediction,the series method combined model is the best,so this model can be used as the National AIDS incidence rate forecast application.