| Objective:Understand the epidemiological characteristics of measles and rubella cases in Changchun from 2009 to 2018;SARIMA model and BP neural network model were used to predict the incidence of measles and rubella in Changchun in 2018.Evaluating the predictive effect of three models so as to obtain the optimal model.These will provide important epidemiological basis for the prevention and control of measles and rubella in Changchun in the future.Methods:The data of measles and rubella in this study were from measles surveillance system,and Changchun's permanent population data were from the statistical yearbook of Jilin Province.This article uses Excel 2016 to integrate and process the data and describe its epidemiological characteristics such as time distribution,regional distribution,age,occupation,and gender;Data were Integrated and processed and described its epidemiological characteristics such as time distribution,regional distribution,age,occupation and gender using Excel 2013;the SARIMA model were established using SPSSS24.0,the BP neural network model were were established using MATLAB2016 a.The above two models were used to predict the incidence of measles and rubella in Changchun in 2018;MAE,RMSE were used to evaluate the model's prediction effect of the two models.Results:1.The incidence of measles was between 0.18/100,000 and 22.03/100,000 from 2009 to 2018,with an average annual incidence of 6.25 / 100,000;the incidence of measles There were obvious seasonal characteristics,mainly in the second quarter,and the proportion of cases was 86.57%.In the population distribution,the cases were mainly small age groups within 2 years old,accounting for 61.25% of the total number of cases,The incidence rate of male was higher than that of female,and the number of scattered children is the largest,accounting for 71.07%,followed by students,accounting for 8.11%.In terms of regional distribution,except for 2017,the incidence of Erdao District in other years was higher than that of other regions.Only one case occurred in Kuancheng district in 2018,no cases occurred in other years.2.The incidence of rubella generally declined from 009 to 2018,from 3.52/100,000 in 2009 to 0.04/100,000 in 2018.There was a small fluctuation in the incidence of rubella in 2014.The rate was 0.37/100,000;in the onset season,it is mainly in the second quarter,accounting for 43.68%;in the population distribution,all disease,of which the highest incidence of cases was 15-39 years old,the proportion is 42.32%,the lowest incidence of cases was <8 months of age,accounting for 3.01%.The incidence rate of male was higher than that of female.The occupational composition was mainly scattered children and students,accounting for 67.02% of all cases;In terms of distribution,the incidence is higher in Chaoyang district and lower in Nong'an county and Nanguan district.3.The corresponding SARIMA model was established based on the monthly incidence of measles from 2009 to 2017,the SARIMA(0,0,1)(0,0,0)12 model is the optional SARIMA model.All model parameters have statistical significance(P<0.05).Using this model to predict the monthly incidence of measles in Changchun in 2018,and the actual values were within the 95% confidence interval of the predicted values.4.The BP neural network model was used the monthly incidence of measles in the previous three years to predict the current monthly incidence.There were a total of 84 samples,the first 60 as the training set,the last 12 as the validation set,and the last 12 as the prediction set.The neural network model was finally determined as 3-6-1 as the final model,using the 3-6-1BP neural network model to predict the incidence of measles in Changchun in 2018.The prediction trend of the model was basically fitted to the actual value,and the prediction accuracy was MAE=0.0276 and RMSE=0.0375.5.In this study,both the SARIMA model and the BP neural network model predicted the monthly incidence of measles in 2018.From the perspective of prediction accuracy,the MAE and RMSE values of the SARIMA model were 0.644 and 1.385 respectively;the MAE and The RMSE values the BP neural network were 0.0276 and 0.0375,respectively.The prediction accuracy shows that the BP neural network model has a better prediction effect.6.The monthly incidence rate of rubella in Changchun from 2009 to 2017 was established as SARIMA model.The SARIMA(3,0,1)(0,0,1)12 model was the optimal SARIMA model for predicting rubella trends in Changchun.The prediction results show that the actual incidence of rubella in Changchun City from January to December 2018 was consistent with the predicted incidence,and was within its 95% confidence interval.The MAE and RMSE values were 0.0493 and 0.0808,respectively.7.The BP neural network model was established for the rubella data using the same method as measles.The optimal BP neural network model for rubella was 3-4-1,using the 3-6-1BP neural network model to predict the incidence of rubella in Changchun in 2018.The actual incidence was within the 95% confidence interval of the predicted value,and the MAE and RMSE values were 0.0095 and 0.0101,respectively.8.From the perspective of prediction accuracy,the MAE and RMSE values of the ARIMA model were 0.0493 and 0.0808.The MAE and RMSE values were 0.0095 and 0.0101,respectively,so the BP neural network model can better predict the monthly incidence of rubella.Conclusions:1.In terms of time distribution,the average annual incidence of measles from 2009 to 2018 is 6.25 per 100,000,and the cases are mainly in the second quarter;in the population distribution,the small age group within 2 years old,mainly scattered children,male incidence Higher;in terms of regional distribution,the incidence rate in Erdao District is higher.2.In terms of time distribution,the incidence of rubella showed a downward trend in 2009-2018,and the cases were mainly in the second quarter;the distribution of the population was mainly in the 15-39 age group,mainly scattered children and students,with a higher incidence of males;In terms of regional distribution,the incidence rate is higher in Chaoyang District.3.Use the monthly incidence of measles and rubella from 2009 to 2017 to establish the SARIMA model and BP neural network model.From the perspective of prediction accuracy,the BP neural network model can better predict the monthly incidence of measles and rubella. |
PDF Full Text Request