Correlation Research Of Hand-foot-mouth Disease And Meteorological Factors In Beijing And Prediction Of Time Series Model

Objective: To grasp the epidemiological characteristics and general situation of hand-foot-mouth disease in Beijing,and explore the correlation and lag effects of certain meteorological factors and the incidence of hand-foot-mouth disease.Construct a time series prediction model suitable for the epidemiological characteristics of HFMD in Beijing to predict its onset trend,and provide a scientific reference for more reasonable formulation of HFMD prevention and control measures suitable for Beijing's climate characteristics.Methods: The software of Excel2013 was used to sort out the incidence data of HFMD in Beijing and the corresponding meteorological data and to establish a time series database.The descriptive epidemiological method was used to characterize and analyze the incidence data of Beijing from 2014 to 2017 and their meteorological factors over the same period.The Spearman correlation analysis method was used to explore the correlation between the number of HFMD cases in Beijing and the meteorological data of each week and the correlation between various meteorological factors.Based on the analysis results,four meteorological indicators and incidence were selected: weekly average temperature,weekly average relative humidity,weekly average pressure,and weekly average wind speed.The data were analyzed by a generalized additive model,and the time series was introduced into the model by natural cubic spline functions in the form of weeks to control the long-term trend and seasonal characteristics of HFMD,and to explore the exposure of various meteorological indicators and the risk of hand-foot-mouth disease Response relationship and its lag effect.Finally,using the incidence data of Beijing from 2014 to 2017 and using SPSS21.0 software to construct a product seasonal ARIMA time series prediction model,and make short-term predictions of the number of cases from 1 to 12 weeks in 2018.The actual incidence data of 1-12 weeks were compared to verify the fitting effect and evaluate the model prediction effect.Result: 1.A total of 127,373 cases of hand,foot and mouth disease were reported in Beijing from 2014 to 2017,of which the outbreak was the worst in 2014,The number of cases in the whole year was 48,011,and the annual incidence rate was 223.1 / 100,000,accounting for about 4 years.37.7% of the number of cases.From 2014 to 2017,there were monthly cases of hand-foot-and-mouth disease in Beijing throughout the year,and the peak time of the disease was probably concentrated in the 18 th to 35 th week of the year,that is,May to August,with the largest number of cases in June.2.Except for the weekly average sunshine hours and the weekly maximum wind speed,the correlations between various meteorological factors and the incidence of HFMD in Beijing were statistically significant.The results of fitting analysis of generalized addition model suggest that the influence of four meteorological indicators such as weekly average temperature(?)on the incidence of HFMD is statistically significant under the condition that the longest lag time is two weeks.In the multi-factor generalized additive model analysis,the meteorological factors lagging by one week have the most significant impact on the incidence of hand foot and mouth disease,the smooth function image indicates that when the weekly average temperature is lower than 17?,the log(RR)value tends to rise as the temperature rises.When it is higher than 17?,the log(RR)value will show a downward trend,and the weekly average relative humidity will gradually increase in the range of [20,38)(%).After that,the log(RR)value will gradually maintain a steady state.3.The SARIMA(2,0,0)(0,1,1)52 seasonal model is a sequence of white noise diagnosed by residual error,Using the actual incidence data of Beijing in 1-12 weeks of 2018 and the model prediction value,the actual number of cases The overall difference from the number of predicted cases is not large,and the prediction accuracy is ideal.Conclusion: 1.The incidence of HFMD in Beijing is seasonal and epidemic.The peak period of the disease is concentrated in the 18-35 weeks of each year,that is,around May to August.The epidemic surveillance and prevention of HFMD should be strengthened during this period.2.Multiple meteorological factors,including the average weekly temperature,are significantly correlated with the incidence of HFMD in Beijing.The results of the GAM fitting analysis suggest that each meteorological factor with a lag period of one week has the most significant impact on the incidence of HFMD in Beijing.3.The seasonal ARIMA prediction model based on weekly time series units can well fit the incidence of HFMD in Beijing and can be applied in the short-term prediction of the incidence of HFMD in Beijing.