The Prediction Model Of The Number Of Brucellosis In China

In 2018,the incidence of brucellosis in the country was 37,947,with an incidence rate of 2.73(1 / 100,000),ranking 8th in the incidence of legal infectious diseases of Class A and B.It is one of the few infectious diseases that continues to rise.The number of spreading areas is also spreading rapidly.It has become one of the social issues that the public is more concerned about.The National Health Commission attaches great importance to it.The prediction of the incidence of brucellosis is necessary for the relevant departments to do a good job of protection in advance.Many scholars have established The prediction model of the number of patients has been introduced,but the common feature of these models is that only the influence of historical data on the incidence of brucellosis is considered,and there is a lack of comprehensive consideration of the factors that affect the incidence.In this context,this article aims to improve the accuracy and rationality of the prediction of the number of Brucellosis patients.Without considering the external factors of the incidence of Brucellosis patients,the ARIMA model is established only from the historical data of the number of Brucellosis patients to predict the number of future patients.On the basis,two main tasks have been done: first,considering the external factors of the incidence of brucellosis,combined with ecological data such as environment,socioeconomics,cattle and sheep production,based on theinfluencing factors studied in the existing literature,a new increase Based on the theoretical assumption that these factors will affect the number of patients,the statistical relationship is analyzed through a scatterplot of the number of patients with brucellosis-influencing factors.The study found that each influencing factor has a linear relationship with the number of patients with brucellosis relationship.Correlation analysis found that the linear relationship between each influencing factor was significant.Therefore,PCA was used to obtain the principal component scores for 15 influencing factors,and the comprehensive principal component scores of influencing factors with lag items and the number of patients with brucellosis were used as the input variables of NARX.The number of patients with brucellosis is an output variable.After repeated trial and error,the network structure when the lag period is 3 and the hidden layer is 13,that is,PCA-NARX(3-13)is an excellent model for the number of patients with brucellosis nationwide from 1995 to 2018.Second,the ARMA(1,1)and PCA-NARX(3-13)two single models,through MSE,MAPE single weight and MAPE-MSE combined weights to build an ARMA-NARX combined model.Through the comparison between the two single prediction models and the combined model obtained from the above research,it is found that in the two single models,PCA-NARX(3-13)has higher prediction accuracy than ARMA(1,1)and so on,which has verified the modeling to a certain extent.The mentioned influencingfactors have a certain impact on the incidence of brucellosis,and the MAPE-MSE combination weighted ARMA-NARX model predicts that the mean square error of the annual incidence of brucellosis in the country is the smallest,because the combined model takes into account external factors and overcomes The shortcomings of each single model.The conclusion shows that the MAPE-MSE combination weighted ARMA-NARX model is a more stable and reasonable model for predicting the number of brucellosis cases in the country,thus providing more real,accurate and reasonable data support for the prevention and control of brucellosis in China.