Epidemiological Characteristics And Trend Prediction Of Scarlet Fever From 2007 To 2016 In Changchun

ObjectivesAim to statistically describe the epidemic situation of scarlet fever in Changchun from 2007 to 2016,and analyze its epidemiological characteristics of different time,regions and populations.At the same time,GM(1,1)model,ARIMA model and BP neural network are used to predict the incidence of scarlet fever of Changchun in 2017.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 scarlet fever in the future of Changchun.MethodsThe data of epidemic monitoring were from the “China Disease Prevention and Control Information System”.The data of permanent population were from the Bureau of Statistics of Changchun.And the 1:100,000 digital map of Changchun was from the College of Earth Surveying and Mapping of Jilin University.The epidemiological analysis of the scarlet fever epidemics was conducted by descriptive epidemiological methods.Data were integrated and processed using Microsoft Excel 2007 software.SPSS 21.0 software was used for statistical analysis of related data.The regional distribution map was drawn using 9.5 software.The epidemic characteristics of scarlet fever in Changchun from 2007 to 2016 were described by the number of cases,incidence and composition ratio.The GM(1,1)model and ARIMA model were established using R software.The BP neural network model was established using Matlab software.To explore the optimal model structure of each model and carry out model test respectively,and predict the incidence of scarlet fever in Changchun City of 2017.Mean absolute error(MAE)and mean squared error(MSE)were used to evaluate the prediction effect of the above three models.Results1.A total of 3,202 cases of scarlet fever were reported from 2007 to 2016 in Changchun City.The annual average incidence rate was 4.26/100,000 and no deaths were reported.The incidence rate in 2011 was 8.58 /100,000,which was the highest in these 10 years.2.The seasonal distribution of scarlet fever in Changchun City from 2007 to 2016 showed a clear “bimodal distribution”.The first peak appears in May-July and the second peak appears in October-December.In terms of regional distribution,the urban incidence rate(6.79 /100,000)was higher than the rural incidence(2.54/00,000),and the high incidence areas were mainly concentrated in Shuangyang District(11.61/ 100,000),the High-tech Development District(10.76/100,000)and Green Park District(9.58/100,000).In terms of the distribution of the population,the incidence and incidence rate of male cases were higher than the female incidence level in the same year.The ratio of male to female morbidity was 1.55:1.The onset ages were mainly concentrated in 0 to 15 years old.Students,scattered children and child care children were the main disease groups,accounting for 41.04%,31.57% and 24.02% of the total number of reported cases,accounting for 96.63% of the total.3.The corresponding GM(1,1)model was established based on the incidence of scarlet fever in the same period of history from 2007 to 2016.The fitting accuracy of the GM(1,1)model established in February and August-December were failed.So the responding extrapolation projections can not be done.The prediction results of incidence in January and March to July were 0.218/100,000,0.088/100,000,0.138/100,000,0.250/100,000,0.370/100,000 and 0.235/100,000.4.Using the time series of incidence of scarlet fever from 2007 to 2016,the ARIMA(2,1,2)model is the optimal ARIMA model for predicting the incidence of scarlet fever in Changchun City.The forecast results showed that the incidence rates of scarlet fever in Changchun from January to December in 2017 were 0.295/100,000,0.194/100,000,0.147/100,000,0.193/100,000,0.283/100,000,0.328/100,000,0.291 /100,000,0.212/100,000,0.169/100,000,and 0.1980.268 /100,000 and 0.309/100,000,respectively.The actual morbidity in each month of 2017 were within the 95% confidence interval of the forecast value.5.In the establishment of BP neural network model,the incidence of the same period of previous three years were used to predict the incidence rate of current month.Considering network performance,prediction effect and simplicity of the model,the BP neural network with 3-9-1 was selected as the final prediction model.The results showed that the incidence of scarlet fever in Changchun from January to December in 2017 were 0.321/100,000,0.145/100,000,0.130/100,000,0.131/100,000,0.135/100,000,0173/10,000 0.334/100,000,0.162/100,000,0.128/100,000,0.128/100,000,0.143/100,000 and 0.199/100,000,respectively.6.For feasibility,ARIMA(2,1,2)models and BP neural networks with a structure of 3-9-1 have successfully predicted the incidence of each month in 2017.The GM(1,1)model only successfully predicted the incidence of January and March-July;For prediction accuracy,the MAE and MSE values of the GM(1,1)model were the smallest,followed by the ARIMA model,and finally the BP model.Conclusion1.A total of 3,202 cases of scarlet fever were reported from 2007 to 2016 in Changchun City.The annual average incidence rate was 4.26/100,000 and no deaths were reported.The incidence rate in 2011 was the highest among these 10 years.2.The scarlet fever in Changchun from 2007 to 2016 presented a clear seasonal distribution with “double peaks”.The two peaks occurred from May to July and from October to December.The incidence rate in urban areas were higher than that in rural areas,in males were higher than that in females.The mainly incidence population were students,scattered children and child care children aged 0 to 15 years.3.Comprehensive comparison of GM(1,1)model,ARIMA model and BP neural network model,ARIMA(2,1,2)model is the best model for predicting the incidence of scarlet fever in Changchun City.