The Epidemiological Characteristics Of Severe Fever With Thrombocytopenia Syndrome And Exploration Of Spatio-temporal Prediction And Early Warning Models In Shandong

Severe fever with thrombocytopenia syndrome(SFTS)is an emerging natural epidemic disease.Due to the lack of preventive measures,the incidence of the disease is increasing in East Asia,Southeast Asia,and North America.Besides the increasing trend of incidence,the high mortality rate of SFTS at the beginning of the epidemic has also triggered research fields and wide public concern.Since the first SFTS patient was confirmed,the distribution of SFTS cases in China has continued to expand.As of 2016,SFTS cases have been reported in 23 provinces across the whole country.Most epidemic areas are located in eastern and central part of China.According to an epidemiological report of SFTS from 2010 to 2016,Shandong Province is the second largest province in China to report SFTS cases,and the number of SFTS cases reported accounting for 26.6%of the total national report.There are few studies on SFTS in Shandong province,however,and many unknown factors affecting SFTS are still in the stage of continuous exploration.This study was based on the direct network reporting system and active surveillance database of Shandong Province from 2011 to 2017,and explores the epidemiological characteristics of SFTS cases in Shandong Province,revealing the differences in the distribution of gender,age,and occupation in spatial-temporal scale;using spatial Bayesian models to explore possible impacts Spatial random effects,individual factors and environmental factors of SFTS fatality rate;based on phase space reconstruction theory and long-term and short-term neural networks to construct a prospective prediction and early warning model of SFTS cases,providing a theoretical basis for the formulation of SFTS prevention and control strategies in Shandong Province and the whole country Technical Support.Main results:1.Epidemiological characteristics:From 2011 to 2017,Shandong Province reported a total of 3178 SFTS cases and 264 deaths,with an average case fatality rate of 8.31%;SFTS cases had a total of 1615 males,150 deaths,and a mortality rate of 9.29%.A total of 1563 females,114 deaths,the mortality rate was 7.29%.The case fatality rate difference between genders was significant(χ2=3.89,P<0.05);SFTS cases were mainly distributed in the age group over 40 years old,and the case occupations were mostly farmer occupations;In the central and eastern coastal areas,there are differences in the gender composition of counties with an annual incidence of more than ten;the SFTS monitoring sequence in Shandong Province has a clear seasonal trend,and the peak of SFTS incidence interval is from May to October every year,and it has a bimodal distribution.2.Factors influencing the case fatality rate:There were spatially cluster trend among death cases of SFTS,and the clustered areas are scattered in the central and eastern Shandong provinces,and their location changes with time;based on the monitoring data modeling from 2011 to 2015,it was found that there was a positive correlation between the precipitation in 2011 and the age factor in 2012 and 2014 with the SFTS mortality rate,and the remaining influencing factors were not statistically significant.3.After predicting the number of SFTS cases in the next 52 weeks,it was found that the long-term and short-term neural network model based on phase space reconstruction has the best extrapolation prediction effect,with RMSE of 5.54 and MAE of 3.46,followed by long-term and short-term neural network models constructed from the original sequence,RMSE of 6.16,MAE of 4.13,and finally ARIMA,RMSE is 6.45,MAE is 5.10.With RMSE as the standard,the accuracy of the model based on phase space reconstruction is improved by 10%and 15%compared to the neural network model and ARIMA model of the original sequence.When using MAE as the standard,the accuracy is improved by 16%and 32%,respectively.Conclusion:1.According to the demographic and spatial-temporal distribution characteristics of SFTS cases of Shandong Province from 2011 to 2017,the implementation of preventive measures should be focused on the middle-aged and elderly farmer groups,and it is necessary to explore the phenomenon that male mortality is significantly higher than female;the implementation of preventive measures should be based on counties and districts with severe SFTS outbreaks in the central and eastern coastal areas of Shandong Province,and intervention measures need to be carried out before the two peak periods from May to October each year.2.The location of the death outcome of SFTS cases has a significant spatial effect and is an important area for the implementation of future prevention strategies;It cannot be inferred that sex was a factor influencing the mortality of SFTS every year during 2011-2015.With the exception of 2012 and 2014,age cannot be inferred in other years as a factor affecting the mortality of SFTS.Except for precipitation in 2011,other environmental factors have not been able to establish a correlation with SFTS mortality.3.Phase space reconstruction based long and short-term neural network models have good prediction performance and can provide technical support for the prospective prediction of SFTS cases in the future.