Spatial-temporal Distribution And Prediction Models Of Mumps In Chongqing

Objectives:This study performed epidemiological analysis based on mumps（mumps）reported incidence data in Chongqing from 2004-2018,spatio-temporal analysis to grasp the disease epidemiological status and spatio-temporal characteristics,and applied Seasonal Autoregressive Intergrated Sliding Evaluation Model（SARIMA）The aim was to provide a scientific basis for mumps prevention and control and to develop prevention and control strategies for the Chongqing authorities.Methods:1.The distribution of mumps in population（gender,age,occupation）,time and area in Chongqing from 2004 to 2018 was analyzed using descriptive epidemiological methods.The year-by-year incidence information of each district and county in Chongqing was imported into the vector map of Chongqing,and ArcGis 10.2 software was used to draw the year-by-year incidence map of Chongqing.2.GeoDa software was used to investigate the spatial dependence of mumps incidence in Chongqing,local spatial incidence patterns and find hotspot areas,mainly including global and local spatial autocorrelation analysis;and pure spatial scan,pure temporal scan and spatio-temporal scan methods were used to analyze the mumps incidence data in Chongqing by SatScan software.The aggregation characteristics of spatial,temporal and spatio-temporal distribution were found.3.R software was used to fit a SARIMA prediction model to the month-by-month mumps incidence data in Chongqing from 2004-2017 and to predict the month-by-month incidence data in 2018;similarly,R software was used to fit a BPNN model to the incidence data from 2004-2017 and to predict the monthly incidence rate in 2018 to compare the model prediction accuracy in order to screen the mumps optimal prediction model.Results:1.There were 158,181 cases of mumps in Chongqing from 2004 to2018,with an average annual reported incidence rate of 36.34/100,000（the lowest was 17.55/100,000 in 2007 and the highest was 59.14/100,000 in2011）.The changing trend of mumps incidence in Chongqing can be divided into three stages:from 2004 down to the lowest value in 2007;then up to the peak in 2011;and finally,from 2011 to 2018 with an overall decreasing trend and a two-year cyclical fluctuation.A total of 93,655 male cases and 65,526 female cases were reported,with an average male to female sex ratio of 1.42:1,which is higher than the male to female sex ratio in the total population of Chongqing.the age group of 5-9 years led the number of patients,accounting for 48.8%of the total number,followed by10-14 years（24.2%）and 0-4 years（15.1%）.In terms of occupational classification,the student population accounted for the largest proportion（62.8%）,followed by kindergarten children（21.3%）and diaspora children（8.0%）.The epidemic of mumps has a clear seasonal pattern,with the first peak in April-July（57.5%）and the second peak in November-January（20.0%）.The annual incidence map of mumps from 2004-2018 shows that the geographical location of the high incidence districts and counties in Chongqing changes from year to year,but is mainly distributed in the west and northeast.The five districts and counties with the highest annual incidence rates were Jiulongpo,Beibei,Nanan,Shapingba,and Dadukou,all located within the main urban area.2.Global autocorrelation analysis in Chongqing from 2004 to 2018found that the spatial distribution of mumps incidence had no spatial dependence within the overall region except for significant positive global spatial correlations in 2004,2007 and 2015.Local spatial autocorrelation analysis identified a total of 16 high-high aggregation areas,14 low-low aggregation areas,13 low-high aggregation areas,and 9 high-low aggregation areas,mainly in the nine districts of the main city,Bishan,Hechuan,Rongchang,Wulong,Wanzhou,and Chengkou.2005,2009,2012,2017,and 2018 did not find incidence"hotspots."Pure spatial scans detected 15 class I aggregation areas and 8 class II aggregation areas with high mumps prevalence.The exact locations of the aggregation areas varied from year to year,but were mainly concentrated in the western,central-western and northeastern parts of Chongqing.The pure time-scan analysis found that the high incidence time aggregation period was from March 2009 to July 2013（LLR=6936.16,P<0.001）.The results of the temporal scan showed that there were four high-incidence mumps aggregation areas in Chongqing from 2004 to 2018,covering 18 districts and counties.The high-incidence spatio-temporal aggregation areas were located in the central-western,western,northeastern,and southwestern parts of Chongqing in order of incidence risk.3.The SARIMA（2,1,1）（1,1,0）₁₂ model was constructed with the month-by-month incidence data of mumps in Chongqing from 2004-2017,and the RMSE between the predicted and actual values in 2018 was 0.9949and the MAPE was 39.8409%.In contrast,the BPNN model predicts an RMSE of 1.1281 and a MAPE of 67.7091%for the 2018 data.In comparison,the ARIMA model predicted better than the BPNN model.CONCLUSIONS:1.Mumps in Chongqing showed a rising and then declining trend,with males,children aged 5-9 years,students,and children in early childhood care being more susceptible to mumps and being a high-risk group that should be focused on prevention and control.The incidence is characterized by a distinct seasonal double-peak distribution,with peaks occurring mainly in April-July and November-January.2.The incidence rate of each district and county has significant spatial heterogeneity and spatio-temporal clustering characteristics,with the high incidence areas mainly distributed in the main city and surrounding districts and counties of Chongqing and the northeast.Relevant health institutions should strengthen real-time spatial surveillance,especially in incidence hotspots and peak periods,improve preventive immunization strategies,and focus on the rational allocation of health resources.3 Compared with the BPNN model,the SARIMA（2,1,1）（1,1,0）₁₂model can better fit the monthly mumps incidence data in Chongqing,with better prediction results and higher prediction accuracy,and can be used for short-term prediction of mumps incidence in Chongqing.