Research On Influenza-Like Cases Prediction Model And Influencing Factors In Urumqi

Influenza is an acute respiratory infectious disease that is easy to spread.Its epidemic poses a serious threat to the global economy and human health.Therefore,the research on the prevention and treatment of influenza is a hot issue at present,and the study of the prevalent trends and early warning models of influenza in the early stage of influenza has important practical and theoretical significance for the prevention and prevention strategies of influenza.Urumqi is located in the northwestern China,central Xinjiang,and the hinterland of the Eurasian continent.It is located in the northern foothills of the North Tianshan Mountains and the southern edge of the Junggar Basin.Urumqi is the world’s farthest city from the ocean.The impact which leads to the meteorological characteristics of the place showing certain specificity,these factors have created powerful conditions for the influenza epidemic in the city.In order to study the relationship between the number of influenza-like illnesses and meteorological factors in Urumqi,this paper collected the number of influenza-like illnesses and meteorological data of the same period(monthly average temperature(℃),monthly precipitation(mm),monthly average pressure(hpa),monthly average relative humidity)from January 2015 to March 2018 in Urumqi(%),Monthly average wind speed(m / s),monthly sunshine hours(h)),these data were subjected to descriptive statistics and Spearman correlation analysis,quantitative analysis of meteorological factors on Urumqi influenza-like illnesses influences.The results show that the number of ILI cases in Urumqi is negatively correlated with monthly average temperature,monthly average wind speed and monthly sunshine hours.The number of ILI cases is positively correlated with monthly average air pressure and monthly average relative humidity.This article uses models to predict and analyze data,so it is very important to choose which model.Because the number of ILI cases is a time series data,and LSTM,ARIMAX,GAM models are all time series methods.Up to now,these three models have been widely used in disease prediction,but no one has used these three models to analyze the meteorological factors affecting the incidence of influenza-like cases in Urumqi and predict the trend of influenza epidemics in Urumqi.So,which model is more suitable for influenza-like cases studies in Urumqi.In order to further quantitatively analyze the impact of meteorological factors on the number of influenza-like cases in Urumqi,we selected LSTM,ARIMAX,and GAM models to study this problem.Therefore,in this study,meteorological factors with strong correlation(| r |> 0.6)were included in these three models,and a model for predicting influenza-like cases in Urumqi was constructed.First,Using the data from January 2015 to September 2017 as the training set,a univariate and multivariable LSTM model was established for the number of influenza-like cases(ILI)in Urumqi,and the data from October 2017 to March 2018 was used as the test set Used for model verification and prediction,estimating model parameters,and selecting the best model from which to compare short-term predictions with real values.Finally,the average absolute percentage error(MAPE)of the model is 27.98%.Secondly,through the residual sequence cross-correlation function(CCF),the monthly average relative humidity and the monthly sunshine hours is correlated with the number of ILI cases.The monthly average relative humidity and monthly sunshine hours were used as influencing variables to establish the ARIMAX model.Among them,the ARIMAX model incorporating the monthly sunshine hours of order 0 lag had the smallest AIC,and all parameters of the model were statistically significant.Finally,according to the prevalence of influenza-like illnesses in Urumqi,Poisson random distribution function is selected as the connection function of the GAM model.The established model uses the AIC minimum principle to select factors and test the goodness of fit,and calculates the regression coefficient β,relative risk(RR)and 95% confidence interval(95% CI)of meteorological factors.This model The MAPE is 12.22%.The results show that the effects of the multivariable Urumqi influenza-like illnesses prediction model established by introducing meteorological factors are better than the univariate prediction model,and the accuracy of the model is improved.Among them,by comparing the methods of the three prediction models,it is concluded that the prediction accuracy of the GAM model with meteorological factors is better,which meets the model’s better grade(10%-20%)in the model evaluation standard.This has important methodological significance and theoretical basis for controlling and preventing the epidemic and spread of influenza in Urumqi.