Time Series Modeling For Fever And Respiratory Syndromic Surveillance In Gansu Province

Objective:To analyze the epidemiological patterns of fever and respiratory syndromes and its pathogens in Gansu province by time series analysis.And to explore the early warning methods for the fever and respiratory syndromic surveillance.We also aimed to provide basis for improving the syndromic surveillance system and improve the ability for strengthening the prevention and controlling of respiratory infectious diseases in Gansu province.Methods:Based on collecting inpatient and outpatient in thirteen sentinel hospitals,we carry out fever and respiratory syndromic surveillance in Gansu province during January 2009 to December 2015.Combining with the case-related demographic information and the pathogenic data of the samples,the distribution of the number of cases of fever respiratory syndrome and the distribution of positive number of pathogens were analyzed by using the observational study and the predictive research method.And through the analysis of the number of cases and the number of positive trends,the number of cases of fever and respiratory syndrome and the number of positive factors occurred in a variety of time series model.Results:1.The incidence of fever and respiratory syndrome in Gansu Province accounted62.91%for men,37.09%for women and 37.65%for children under the age;and the peaks were in March and November to December in each year.The positive rate of viruses was 23.29%and adolescents and older people was higher.It had obvious seasonal character,which reached the peaks in March and November to December,and 13.79%detection rate of influenza viruses was highest and 8.63%detection rate of rhinovirus was followed.2.While the detection rate of bacterial was32.83%and 43.38%detection rate of older people was highest.With the increase of age,the detection rate of bacterial increased（Z=-11.89,P<0.01）and male was significantly higher than female（χ²=6.174,P=0.013）.3.ARMA（1,1）model was suitable for the fever and respiratory syndrome in Gansu Province,which expressed asY_t-82.66=0.48(Y_t-1-82.66)+ε_t（10）0.34ε_t-1.ARIMA（1,1,1）（0,1,1）12 model fitted the pathogen positive sequence information,with a good fitting effect,which formula is（1-0.40 B）(1-B¹²)（1-B）（Y_t（10）0.22）（28）（1-0.87 B）(1-0.63B ¹²)ε_t.4.The VARX（1,0）model,which contains the sequence of the fever respiratory syndrome and the positive sequence of the pathogen,isY_t（28）0.36Y_t-1（10）0.19X_t（10）ε_t.Conclusions:1.In Gansu Province,either the distribution of fever and respiratory syndrome or pathogen-positive series have bimodal seasonal distribution characteristics,which the peaks in March and November to December in each year.ARIMA model can fully extract the information in the sequence of fever and respiratory syndrome in Gansu Province,and the fitting effect is good,but it can not reflect the relationship between the syndrome number and the pathogen positive number.2.VARX（1,0）model fully embodies the corresponding relationship between the sequence of fever and respiratory syndrome and pathogen positive sequence in Gansu Province.Based on the mathematical views,the model confirmed the significance of syndrome monitoring to disease prevention and control,also reflected the syndromic surveillance and diagnosis of disease is indispensable for early warning.