Research And Application Of Influenza Prewarning Technology In Qinghai Plateau

Objective:To establish an influenza prewarning system suitable for Qinghai Province,and use the influenza prewarning system to determine the influenza epidemic threshold,predict the influenza epidemic status,and understand the effects of various meteorological factors on the influenza epidemic,so as to provide a theoretical basis for influenza prevention and control in Qinghai Province.Methods:1.The influenza surveillance data of Qinghai Province from 14 th week of 2013 to the 13 th week of 2020 was obtained through the "China Influenza Surveillance Information System",and organize the data according to the format required for the study.The moving epidemic method was used to process the positive rate data of influenza virus detection from 14 th week of 2013 to the 13 th week of 2019,obtain the influenza epidemic threshold and graded intensity threshold;the model effect was evaluated through parameters such as sensitivity,specificity,and Youden index;and then the obtained influenza epidemic thresholds and graded intensity thresholds was used to evaluate the influenza epidemic intensity in Qinghai Province from the 14 th week of 2019 to the 13 th week of 2020.2.The meteorological data was obtained through the China Meteorological Data Sharing website,and organized in the format required for the study.The Long Short-Term Memory model was utilized to fit the meteorological data and influenza virus detection positive rate data from the 14 th week of 2013 to the 13 th week of 2019 to establish a model.The model was used to predict the influenza virus detection positive rate from the 14 th week of 2019 to the 13 th week of 2020,and the root mean square error was obtained to verify the effect of the model.3.The data was processed through correlation analysis and collinearity analysis to determine the meteorological factors included in the model.The data on the detection positive rate of influenza virus detection and the data on influenza-like cases were fitted with meteorological data,respectively,and a nonlinear model with distribution lag was established.The obtained DLNM models were compared to determine the optimal model,and the optimal model was used to determine the influence and delay effect of various meteorological factors on influenza epidemics.Results:1.The corresponding influenza epidemic threshold and graded intensity threshold were obtained through the MEM model.The epidemic outbreak threshold,epidemic end threshold,medium intensity threshold,high intensity threshold and extremely high intensity threshold were 16%,7%,29%,48% and 60%,respectively.The sensitivity of this study was 93%,the specificity was 96%,and the model fitting effect was good.2.The LSTM model was established.After training,the loss function was reduced to 0.05,the RMSE was 0.126,and the prediction effect was relatively good.3.In this study,the result of multivariate DLNM model of influenza-like cases was good.Higher and lower temperatures increased the relative risk of influenza at a delay of 0-1 week.Low mean relative humidity increased the relative risk of influenza at delays of 1 and 4 weeks.Various wind speeds increased the relative risk of influenza at a delay of 1-2 weeks.Shorter sunshine hours increased the relative risk of influenza at a delay of 2 weeks.Longer sunshine hours increased the relative risk of influenza at a delay of 0 week.Higher precipitation increased the relative risk of influenza at a delay of 4 weeks.Conclusion:This study established an influenza prewarning system suitable for Qinghai Province.The MEM model established in this study has high sensitivity and specificity in Qinghai Province.The influenza epidemic threshold obtained by the MEM model can be applied to the whole Qinghai Province,with strong suitability.The LSTM model established in this study has a good effect and can accurately predict the influenza epidemic status of cities and states in Qinghai Province.This study showed that there is a relationship between meteorological factors and the spread of influenza.Lower average temperature,higher average temperature,larger temperature difference,and lower average relative humidity result in an increase in the relative risk of influenza.