多源数据在呼吸道传染病预警中的应用效果分析

Analysis on the effect of multi-source data in early warning of respiratory infectious diseases

目的 探讨多源数据在呼吸道传染病预警中的作用,为呼吸道传染病的预警提供参考。方法 选取2022年1月1日-2023年5月31日山东省潍坊市呼吸道传染病预警多源数据,以潍坊市呼吸道传染病数据为参照数据,运用交叉相关函数与门诊主诉、初步诊断、120急救、百度药物、百度发热、百度症状等数据进行滞后相关性分析,并运用EARS-3Cs模型构建预警模型。结果 百度症状检索数据领先于潍坊市呼吸道传染病数据2 d,且滞后相关系数(ACF)为0.749;百度药物、百度发热、初步诊断数据滞后天数为0,其中百度药物的ACF最大(0.885);门诊主诉滞后1 d, 120数据滞后10 d。按照多源数据滞后天数<0且滞后相关系数最大的原则,选取百度症状数据构建EARS-3Cs模型,其C1、C2、C3值超过预警阈值的天数分别是3、10、24 d,且主要集中在2022年12月、2023年3月。结论 滞后相关性分析可从多源数据中找到最优数据源,EARS-3Cs模型对呼吸道传染病预警有一定的可靠性和合理性。

Objective To explore the role of multi-source data in respiratory infectious disease early warning,so as to provide reference for respiratory infectious disease early warning.Methods The multi-source data of early warning of respiratory infectious diseases in Weifang city,Shandong province from January 1,2022 to May 31,2023 were selected.Based on the data of respiratory infectious diseases in Weifang city,the cross-correlation function was used to analyze the lag correlation with the data of outpatient complaints,preliminary diagnosis,120 first aid,Baidu drugs,Baidu fever,Baidu symptoms,and the EARS-3Cs model was used to construct an early warning model.Results The symptom retrieval data of Baidu was 2 days ahead of the data of respiratory infectious diseases in Weifang city,and the lag correlation coefficient(ACF)was 0.749.The lag days of Baidu drug,Baidu fever and preliminary diagnosis data were 0,and the ACF of Baidu drug was the largest(0.885).The outpatient complaints lagged by 1 day,and the data of 120 lagged by 10 days.According to the principle that the lag days of multi-source data were<0 and the lag correlation coefficient was the largest,Baidu symptom data was selected to construct the EARS-3Cs model,and the days when the C1,C2,and C3 values exceeded the early warning threshold were 3,10,and 24 days,respectively,and were mainly concentrated in December 2022 and March 2023.Conclusion Lag correlation analysis effectively identified the optimal data source among multi-source data,and the EARS-3Cs model exhibited reasonable reliability and validity for early warning of respiratory infectious diseases.