重庆市2009—2016年细菌性痢疾空间流行病学特征及基于气象要素的预测模型研究
|
摘要
目的·分析重庆市细菌性痢疾的空间流行病学特征及其与气象要素的相关性,并构建其发病率预测模型,为重庆市细菌性痢疾疫情的防控提供科学依据。方法·收集2009—2016年重庆市细菌性痢疾及气象要素数据,并进行描述性流行病学分析,采用时空扫描统计量进行细菌性痢疾时空聚集性分析,运用DCCA系数法量化细菌性痢疾发病率与气象要素的相关性,运用Boruta算法结合粒子群优化算法(particle swarm optimization,PSO)及支持向量机回归模型(support vector machine for regression,SVR)构建细菌性痢疾发病率预测模型。结果·①2009—2016年重庆市细菌性痢疾年均报告发病率为29.394/100 000,0～5岁年龄组发病率(295.892/100 000)最高,散居儿童占比(50.335%)最大,5月—10月为其季节性发病高峰;细菌性痢疾呈现显著的时空聚集性,一类聚集区主要集中在重庆市主城区,二类聚集区主要集中在重庆市东北地区;6月—10月为其主要的聚集时间。②与人群细菌性痢疾发病率具有很强相关性的气象要素分别为月平均气压(ρ_(DCCA)=-0.918)、月平均最高气温(ρ_(DCCA)=0.875)及月平均气温(ρ_(DCCA)=0.870)。③基于气象要素构建的PSO_SVR模型均方误差(mean squared error,MSE)、平均绝对百分比误差(mean absolute percentage error,MAPE)、平方相关系数(square correlation coefficient,R2)分别为0.055、0.101及0.909。结论·重庆市主城区及渝东北地区应作为细菌性痢疾的重点防控区域,同时相关卫生部门应结合气象要素与细菌性痢疾发病率的密切相关性及其季节性高发特点,对0～5岁儿童、散居儿童、农民等人群采取针对性的应对措施以控制细菌性痢疾传播与流行。基于气象要素建立的PSO_SVR模型预测性能良好,可为细菌性痢疾的防控提供有力的理论支撑。
Objective · To analyze the spatial epidemiological characteristics of bacillary dysentery and its correlation with meteorological elements in Chongqing, and to construct its incidence prediction model, thus providing scientific basis for the prevention and control of bacterial dysentery. Methods · The data of bacterial dysentery cases and meteorological factors from 2009 to 2016 in Chongqing was collected in this study. Descriptive methods were employed to investigate the epidemiological distribution of bacillary dysentery. Spatiotemporal scanning statistics was used to analyze spatiotemporal characteristics of bacillary dysentery. DCCA coefficient method was used to quantify the correlation between the incidence of bacillary dysentery and meteorological elements. Both Boruta algorithm and particle swarm optimization algorithm(PSO) combined with support vector machine for regression model(SVR) were used to establish the prediction model for the incidence of bacterial dysentery. Results · ① The mean annual reported incidence of bacillary dysentery in Chongqing from 2009 to 2016 was 29.394/100 000. Children <5 years old had the highest incidence(295.892/100 000) among all age categories and scattered children had the highest proportion(50.335%) among all occupation categories. The seasonal incidence peak was from May to October. Bacterial dysentery showed a significant spatialtemporal aggregation that the most likely clusters for disease was found mainly in the main urban areas and main gathering time was from June to October. ② The most important meteorological elements associated with the incidence of bacterial dysentery were monthly mean atmospheric pressure(ρ_(DCCA)=-0.918), monthly mean maximum temperature(ρ_(DCCA)=0.875) and monthly mean temperature(ρ_(DCCA)=0.870). ③ The mean squared error(MSE), mean absolute percentage error(MAPE) and square correlation coefficient(R2) of PSO_SVR model constructed based on meteorological elements were 0.055, 0.101 and 0.909, respectively.Conclusion · The main urban areas of Chongqing and the northeast of Chongqing should be regarded as the key areas for the prevention and control of bacillary dysentery. At the same time, according to the characteristics of bacillary dysentery, relevant health departments should take targeted measures to control the spread and prevalence of bacillary dysentery among children <5 years old, scattered children and farmers. The PSO_SVR model constructed based on meteorological elements has good predictive performance and can provide scientific theoretical support for the prevention and control of bacterial dysentery.
Objective · To analyze the spatial epidemiological characteristics of bacillary dysentery and its correlation with meteorological elements in Chongqing, and to construct its incidence prediction model, thus providing scientific basis for the prevention and control of bacterial dysentery. Methods · The data of bacterial dysentery cases and meteorological factors from 2009 to 2016 in Chongqing was collected in this study. Descriptive methods were employed to investigate the epidemiological distribution of bacillary dysentery. Spatiotemporal scanning statistics was used to analyze spatiotemporal characteristics of bacillary dysentery. DCCA coefficient method was used to quantify the correlation between the incidence of bacillary dysentery and meteorological elements. Both Boruta algorithm and particle swarm optimization algorithm(PSO) combined with support vector machine for regression model(SVR) were used to establish the prediction model for the incidence of bacterial dysentery. Results · ① The mean annual reported incidence of bacillary dysentery in Chongqing from 2009 to 2016 was 29.394/100 000. Children <5 years old had the highest incidence(295.892/100 000) among all age categories and scattered children had the highest proportion(50.335%) among all occupation categories. The seasonal incidence peak was from May to October. Bacterial dysentery showed a significant spatialtemporal aggregation that the most likely clusters for disease was found mainly in the main urban areas and main gathering time was from June to October. ② The most important meteorological elements associated with the incidence of bacterial dysentery were monthly mean atmospheric pressure(ρ_(DCCA)=-0.918), monthly mean maximum temperature(ρ_(DCCA)=0.875) and monthly mean temperature(ρ_(DCCA)=0.870). ③ The mean squared error(MSE), mean absolute percentage error(MAPE) and square correlation coefficient(R2) of PSO_SVR model constructed based on meteorological elements were 0.055, 0.101 and 0.909, respectively.Conclusion · The main urban areas of Chongqing and the northeast of Chongqing should be regarded as the key areas for the prevention and control of bacillary dysentery. At the same time, according to the characteristics of bacillary dysentery, relevant health departments should take targeted measures to control the spread and prevalence of bacillary dysentery among children <5 years old, scattered children and farmers. The PSO_SVR model constructed based on meteorological elements has good predictive performance and can provide scientific theoretical support for the prevention and control of bacterial dysentery.
引文
[1]张楠,曾德唯,方勇.重庆市南岸区2011年细菌性痢疾报告及监测结果分析[J].现代预防医学,2013,40(19):3694-3695,3699.
[2]常昭瑞,孙强正,裴迎新,等.2012年中国大陆地区细菌性痢疾疫情特点与监测结果分析[J].疾病监测,2014,29(7):528-532.
[3]罗玲,黎明.湖北省水痘发病情况与气象因素的关系研究[J].中国全科医学,2018,21(9):1114-1117.
[4]栾博.吉林省2008~2016年手足口病流行特征分析及时间序列模型的应用[D].长春:吉林大学,2017.
[5]陈虹汝,周亮,李高明,等.基于气象因素的手足口病发病风险预测模型[J].第三军医大学学报,2017,39(12):1292-1297.
[6]Zebende GF.DCCA cross-correlation coefficient:quantifying level of crosscorrelation[J].Phys Stat Mech Appl,2011,390(4):614-618.
[7]Kulldorff M,Heffernan R,Hartman J,et al.A space-time permutation scan statistic for disease outbreak detection[J].PLoS Med,2005,2(3):e59.
[8]Vassoler RT,Zebende GF.DCCA cross-correlation coefficient apply in time series of air temperature and air relative humidity[J].Phys Stat Mech Appl,2012,391(7):2438-2443.
[9]Kursa MB,Rudnicki WR.Feature selection with the boruta package[J].J Stat Softw,2010,36(11):1-13.
[10]Poli R,Kennedy J,Blackwell T.Particle swarm optimization[J].Swarm Intell,2007,1(1):33-57.
[11]Cao J,Cui H,Shi H,et al.Big data:a parallel particle swarm optimizationback-propagation neural network algorithm based on mapreduce[J].PLoS One,2016,11(6):e0157551.
[12]Podobnik B,Grosse I,Horvati?D,et al.Quantifying cross-correlations using local and global detrending approaches[J].Eur Phys J,2009,71(2):243-250.
[13]李钰曼.改进的PSO-RBF神经网络在复杂工业过程中的应用[D].石家庄:河北科技大学,2018.
[14]Chang Z,Zhang J,Lu R,et al.The changing epidemiology of bacillary dysentery and characteristics of antimicrobial resistance of Shigella isolated in China from 2004-2014[J].BMC Infect Dis,2016,16(1):685.
[15]Esmaeily H,Tayefi M,Ghayour-Mobarhan M,et al.Comparing three data mining algorithms for identifying the associated risk factors of type 2diabetes[J].Iran Biomed J,2018,1(4):133-134.
[16]Zhang J,Nawata K.A comparative study on predicting influenza outbreaks[J].Biosci Trends,2017,11(5):533-541.
[17]孙小宇,初艳慧,张震,等.3种预测模型在北京市西城区痢疾发病数预测中的应用与比较[J].现代预防医学,2014,41(19):3470-3474.
[18]郑慧敏,薛允莲,黄燕飞,等.ARIMA模型在深圳市法定传染病发病趋势预测的应用[J].实用预防医学,2016,23(02):240-243.
[19]解合川,任钦,曾海燕,等.支持向量机在传染病发病率预测中的应用[J].现代预防医学,2013,40(22):4105-4108,4112.
[2]常昭瑞,孙强正,裴迎新,等.2012年中国大陆地区细菌性痢疾疫情特点与监测结果分析[J].疾病监测,2014,29(7):528-532.
[3]罗玲,黎明.湖北省水痘发病情况与气象因素的关系研究[J].中国全科医学,2018,21(9):1114-1117.
[4]栾博.吉林省2008~2016年手足口病流行特征分析及时间序列模型的应用[D].长春:吉林大学,2017.
[5]陈虹汝,周亮,李高明,等.基于气象因素的手足口病发病风险预测模型[J].第三军医大学学报,2017,39(12):1292-1297.
[6]Zebende GF.DCCA cross-correlation coefficient:quantifying level of crosscorrelation[J].Phys Stat Mech Appl,2011,390(4):614-618.
[7]Kulldorff M,Heffernan R,Hartman J,et al.A space-time permutation scan statistic for disease outbreak detection[J].PLoS Med,2005,2(3):e59.
[8]Vassoler RT,Zebende GF.DCCA cross-correlation coefficient apply in time series of air temperature and air relative humidity[J].Phys Stat Mech Appl,2012,391(7):2438-2443.
[9]Kursa MB,Rudnicki WR.Feature selection with the boruta package[J].J Stat Softw,2010,36(11):1-13.
[10]Poli R,Kennedy J,Blackwell T.Particle swarm optimization[J].Swarm Intell,2007,1(1):33-57.
[11]Cao J,Cui H,Shi H,et al.Big data:a parallel particle swarm optimizationback-propagation neural network algorithm based on mapreduce[J].PLoS One,2016,11(6):e0157551.
[12]Podobnik B,Grosse I,Horvati?D,et al.Quantifying cross-correlations using local and global detrending approaches[J].Eur Phys J,2009,71(2):243-250.
[13]李钰曼.改进的PSO-RBF神经网络在复杂工业过程中的应用[D].石家庄:河北科技大学,2018.
[14]Chang Z,Zhang J,Lu R,et al.The changing epidemiology of bacillary dysentery and characteristics of antimicrobial resistance of Shigella isolated in China from 2004-2014[J].BMC Infect Dis,2016,16(1):685.
[15]Esmaeily H,Tayefi M,Ghayour-Mobarhan M,et al.Comparing three data mining algorithms for identifying the associated risk factors of type 2diabetes[J].Iran Biomed J,2018,1(4):133-134.
[16]Zhang J,Nawata K.A comparative study on predicting influenza outbreaks[J].Biosci Trends,2017,11(5):533-541.
[17]孙小宇,初艳慧,张震,等.3种预测模型在北京市西城区痢疾发病数预测中的应用与比较[J].现代预防医学,2014,41(19):3470-3474.
[18]郑慧敏,薛允莲,黄燕飞,等.ARIMA模型在深圳市法定传染病发病趋势预测的应用[J].实用预防医学,2016,23(02):240-243.
[19]解合川,任钦,曾海燕,等.支持向量机在传染病发病率预测中的应用[J].现代预防医学,2013,40(22):4105-4108,4112.