Aberration Detection,Prediction And Early Warning Of Hand,Foot And Mouth Disease In The Pearl River Delta,China

BACKGROUND:Hand-foot-and-mouth disease(HFMD)is a common and highly contagious childhood infectious disease caused by a variety of enteroviruses and has become one of the key infectious diseases that endanger children's health.Since 2010,number of HFMD cases in Guangdong Province has been ranked first in our country,and the incidence rate is also the top five in our country.With the grow of Chinese total economy and the advancement of urbanization,the migration of the population has become frequently,resulting in increasing HFMD morbidity in Guangdong Province.In view of HFMD epidemic had adverse impact on health of children,it is important to understand the epidemiological characteristics and related influencing factors of the disease,carry out research of disease outbreak detection and early warning,and explore new methods and strategies for prevention and control of HFMDOBJECTIVE:We aim to reveal epidemiological characteristics of HFMD,to establish sensitive models of HFMD for outbreak detection,to analyze related incidence factors of HFMD,to establish early-warning model by using associated factors and conduct prediction and early warning of the incidence of HFMD through the model.METHODS:We collected HFMD data,meteorological data,social and demographic data in nine cities of Pearl River Delta region from 2011 to 2017.Construction data of time series on HFMD and various meteorological variables,we analyzed the incidence,severe and death cases of HFMD;and descriptive analysis was used to analyze the population distribution,temporal,spatial and spatial-temporal distribution of HFMD cases.This study used the FarringtonFlexible and EARSC to identify aberrances in weekly counts of HFMD in cities of Pearl River Delta region during 2017.We used Spearman correlation analysis to analyze the relationship between the meteorological factors and the HFMD case;and Distributed Lag Non-linear Model was applied to estimate the association between temperature,rainfall,relative humidity and morbidity for HFMD.Support Vector Machine,Generalized additive model and Multivariate linear regression model was applied to predict the weekly counts of HFMD in cities of Pearl River Delta region during week 22 to 52 of 2017.Root Mean Square Error and R-square were used to assess the performance and the goodness of fit of the models.Finally,according to the optimal model for HFMD early prediction and warning.RESULTS:1.In Pearl River Delta region,1902611 cases,including 2992 severe cases and 82 death cases,were reported from 2011 to 2017 and the incidence was 465.9/100000.Among them,male cases were 1165617(61.3%),female cases were 736994(38.7%),and the sex ratio was 1.58.The cases of HFMD occurred mostly in children under 4 years old(80.9%),especially in children under 1 years old(600019,31.5%).The cases of HFMD occurred mostly in scattered children(1468278,77.2%).2.There was a seasonal pattern for case count in Pearl River Delta region.Two peaks often occurred in April to July and September to November.The high incidence of HFMD is mainly concentrated in Zhuhai,Foshan,Zhongshan and Huizhou,with the largest incidence in Zhuhai(more than 550/100000)and the largest cases in Guangzhou.3.The model constructed by FarringtonFlexible can accurately and sensitively detect the outbreaks of HFMD during epidemic period in the Pearl River Delta region.4.The correlation coefficient between weekly average temperature,weekly relative humidity,weekly rainfall,and weekly HFMD cases were 0.8,0.4 and 0.6,respectively.We found temperature,rainfall and relative humidity were associated with HFMD by using Distributed Lag Non-linear Model5.The root mean-square error(RMSE)of SVR model was 0.06,the GAM was 0.09,and multivariate linear regression model was 0.11,respectively.The R-squared of three models were 0.92 for SVR model,0.75 for GAM,0.73 for multivariate linear regression modelCONCLUSIONS:1.The incidence of HFMD increased in the Pearl River Delta from 2011 to 2017,while severe and death cases showed a downward trend.The incidence of HFMD had a seasonal pattern and the peak is mainly concentrated in May to June and October2.The model constructed with the FarringtonFlexible method can be applied to the detection of HFMD outbreaks3.Temperature,rainfall,relative humidity,severe cases of HFMD were closely related to the incidence of HFMD.The model established by using these indicators can better predict the early onset of foot and mouth disease4.Support vector machine has higher accuracy and less deviation,and it was feasible and effective to use it to predict the incidence of HFMD in the Pearl River Delta region.