Prediction Of The Trend Of Hand-foot-mouth Disease Based On Big Data

Hand,foot,and mouth disease(HFMD)is a common infectious disease caused by enteroviruses,and it is becoming more and more popular in China.Its incidence is obviously regional and seasonal,and previous studies have reported the impact of meteorological factors on the incidence of HFMD.There are also many models that predict the trend of HFMD.It is of great significance that the prediction of its incidence trend can help relevant institutions to formulate preventive measures.In this paper,based on the multivariate and heterogeneous big data of HFMD,the characteristics of high correlation were obtained through correlation analysis,and combining with LSTM network the data of HFMD in jinan and guangzhou were modeled,so as to make mid-term prediction of the incidence trend.In other prediction analysis,one-step prediction is generally adopted and other features are not considered.However,the research problem in this paper is to predict the incidence of HFMD for a long time.The difficulty in this paper is how to ensure the accuracy of the prediction for a long time and how to effectively use other features.In view of the above problems,this paper chooses meteorological factors as the characteristics to be used,first of all to study the impact of meteorological factors on the incidence of HFMD.After analyzing the epidemiological and climatic characteristics of HFMD in the two cities,Spearman correlation analysis was used to explore the correlation between various meteorological factors and HFMD,choose high correlation between meteorological factors and eliminate collinearity,on the basis of the use of distributed lag non-linear model to estimate nonlinear effect of meteorological factors on the HFMD.The results show that temperature and relative humidity have an effect on the incidence of HFMD,and this effect has spatial heterogeneity.In jinan,the risk of disease increases with the increase of temperature,and the risk of disease is greater in the case of low humidity,and it has the effect of lagging about 10 days.When the temperature in guangzhou is 27 ℃,the risk of disease is the highest.With the increase of humidity,the risk of disease will be reduced,and it will be affected within 1-2 weeks.After determining the use temperature and relative humidity as characteristics,Long-short Term Memory(LSTM)is used for modeling and prediction.Firstly,the prediction method is introduced in detail,and then the network structure,training process,and parameter tuning of the model are introduced.The prediction results of the LSTM Model were compared with the use of Seasonal Autoregressive Integrated Moving Average Model(SARIMA),Support Vector Regression(SVR)and CNN-LSTM models.Except for the jinan data set,the SARIMA Model was slightly better than the LSTM Model.Others are the Root Mean Squared Error(RMSE)and Mean Absolute Error(MAE)of the LSTM model are smaller,indicating that the prediction accuracy of the model is higher.In this paper,temperature and humidity were incorporated into the LSTM model as variables,and the prediction accuracy was improved,and the results of the two data sets were better than those of other models.The results of this study show that the LSTM model has a good prediction accuracy in the prediction of the incidence trend of HFMD.The average temperature and relative humidity affect the transmission of HFMD.The establishment of the model based on temperature and humidity can improve the prediction accuracy.Modeling combined with big data has a good application prospect in the prediction of HFMD.