Spatio-temporal Data Analysis Model And Its Application In The Prediction Of Hand,Foot And Mouth Disease

With the advancement of science and technology,the scope of human exploration of the world is constantly expanding,and a large amount of spatio-temporal data is generated.Spatio-temporal data refers to data that has a time dimension feature and a spatial dimension feature,which usually presents spatio-temporal non-stationary and the disease data is one of them.At present,more and more attention has been paid to epidemiology based on spatiotemporal data analysis and modeling.Hand,foot and mouth disease(HFMD)is an infectious disease distributed throughout the world,especially in Asia,which ranks first in the national statutory category C infectious diseases and causes serious harm to many families.Based on exploratory data analysis,this paper makes research from the following two aspects.A time series analysis model was established to predict the short-term trend of HFMD considering only the influence of time factors on disease transmission;Second,Temporally Weighted Regression(TWR),Geographically Weighted Regression(GWR),and Geographically and Temporally Weighted Regression(GTWR)were established respectively to explore the relationship between HFMD epidemic trend and its risk factors(meteorological,demographic and economic,etc.)with the consideration of Time,space,and its risk factors.First,the exploratory data analysis method was used to analyze the time,space,epidemiological characteristics,and Temporal and Spatio clustering of HFMD.The results showed that the incidence rate of HFMD cases was decreased,and the onset age was mainly under the age of 9 years,accounting for 96.126%of the total,among which scattered children accounted for the majority,while the incidence of male children was higher than female children.The main virus causing HFMD was EV71,and there is obvious spatial clustering in the pathogenesis of HFMD.Secondly,based on the exploratory data analysis results,the article took the incidence data of children aged under 9 years in 2009 from 102 counties in the Inner Mongolia Autonomous Region as samples,and established a seasonal autoregressive integrated moving average model with weekly cycle and monthly cycle respectively.To predict the overall incidence of HFMD and compare the forecast results,and to provide early warning for HFMD.Finally,we introduce the GTWR model based on the inadequacies that time series analysis was easy to be interfered by external environment to affect prediction results.In order to test its performance,10 factors that may be related to the development of HFMD were selected to establish the OLS,TWR,GWR,and GTWR models.The spatially non-stationary test of the models and the regression coefficients was presented in the form of a graph through visualization techniques.The result showed that the accuracy of prediction was with the consideration about the non-stationary of space and time.The GTWR model(0.495)has a better degree of goodness of fit than the GWR model(0.367)and the TWR model(0.113).The GTWR model could better reflect the spatial relationship between the incidence rate and the influencing factors.The improvement of the GTWR model was statistically significant.