Study On The Epidemic Characteristics Of Tuberculosis And The Model Prediction Of The Number Of Cases In Hefei

Objective:To analyze the surveillance data of tuberculosis epidemic situation in Hefei from 2009 to 2020,and describe the distribution and epidemiological characteristics of patients in different time,region and population.Establish a prediction model of tuberculosis cases,predict the future incidence trend,and provide a reference for the establishment of the incidence prediction system.From the spatial dimension,the clustering characteristics of tuberculosis were statistically analyzed using spatial analysis software.To provide scientific basis for tuberculosis epidemic prevention and control in Hefei.Methods:42681 cases of pulmonary tuberculosis patients were screened out by collecting the data of tuberculosis in Hefei from 2009 to 2020 in the subsystem"TB Management Information System"of"China Disease Prevention and Control Information System".Geo Da and Arcgis 10.8 spatial analysis software were used for analysis.Spatial distribution maps were drawn at the street/township level to explore the spatial distribution rules and cold and hot spots of tuberculosis incidence.The ARIMA and Holt-Winters index smoothing models were constructed to predict tuberculosis data in Hefei by using the monthly reported cases.Result:（1）A total of 42681 cases of pulmonary tuberculosis were reported in Hefei from 2009to 2020.The incidence rate decreased from 57.96/100000 in 2009 to 31.04/100000 in2020,showing a fluctuating downward trend(χ_trends²=12.53,P<0.001);The sex ratio of men and women is 2.55:1.The incidence of pulmonary tuberculosis in Hefei is characterized by obvious periodicity and seasonality,with a trough in February,a big peak in March-April,and a small peak in November-December.The age is mainly concentrated in the population aged 20～and over 70,with the majority of farmers,domestic workers and unemployed.Feidong in the region shows a gradual downward trend,and Luyang District has been at a low reporting level.（2）The global spatial autocorrelation results showed that the incidence of tuberculosis in Hefei from 2009 to 2014 presented spatial autocorrelation（Moran’s I values were all>0,P values were all<0.05）,The spatial distribution of 2015—2020 is stochastic.The local autocorrelation analysis showed that the high-high incidence streets/towns were mainly concentrated in Changfeng County and Feidong County;Low-low concentration areas are mainly concentrated in the main urban areas（Luyang District and Baohe District）.According to the analysis of cold and hot spots,there are two hot spots,distributed in Yijing Township,Changfeng County and Yandian Township,Feixi County.（3）ARIMA and Holt-Winters additive index models have good fitting effects on the monthly reported cases of tuberculosis in Hefei from 2009 to 2020.The optimal model ARIMA（0,1,1）（0,1,1）₁₂（AIC=1255.58）is selected according to the principle of minimum AIC,and its residual is（χ²=19.61,P=0.07）through Box-Ljung test,indicating that the residual is white noise,which is sufficient for data extraction;The goodness of fit of Holt-Winters additive exponential model is 0.74.The model fitting is good,and the difference of Ljung-Box residual test results is not statistically significant（Q=32.28,P=0.07）,indicating that the residual of the fitting results is white noise with random distribution,and the model fitting is good.Compared with ARIMA（0,1,1）（0,1,1）₁₂（mean absolute percentage error 10.03,root-mean-square error 41.08）,the Holt-Winters additive exponential model has better goodness of fit（mean absolute percentage error 9.99,root-mean-square error 39.05）;The two models were used to predict the data in 2020 and compared with the actual values.It was found that the predicted values of the two models were the least in February 2020,and the relative errors of ARIMA（0,1,1）（0,1,1）₁₂and Holt-Winters additive index models for February fitting were also the largest,with values of 0.38 and 0.41 respectively.Conclusions:The epidemic of tuberculosis in Hefei has a certain degree of aggregation.In the follow-up,different prevention and control measures should be taken according to the characteristics of different counties（districts）,streets/towns,and the prevention and control efforts should be strengthened in the streets and hot spots in the high-high concentration areas,focusing on young and middle-aged men and farmers,and increasing publicity efforts during the Spring Festival and before the seasonal exchange to improve the awareness rate and self-protection level of the people.ARIMA model and Holt-Winters index smoothing model can be used to predict the reported incidence of tuberculosis in Hefei,which can provide data reference for relevant departments.