Bootstrap在流感监测预警中的应用

发布时间：2018-02-26 03:23

  本文关键词： 流感样病例 bootstrap 移动平均法 预警　出处：《南京医科大学学报(自然科学版)》2017年10期 　论文类型：期刊论文

【摘要】：目的:应用bootstrap方法估计流感样病例(influenza-like illnesses,ILI)发病水平95%置信区间(95%CI)作为预警阈值,评价该方法在ILI预警中的价值。方法:以2005—2014年(除2006、2009年)的周监测数据为基线,采用历史2年流感病毒最高阳性检出率的30%作为流感流行标准,确定2006年和2009年流感流行期金标准,通过回顾性分析,比较bootstrap法、移动平均法和金标流行期的一致性;并以2012—2015年的周监测数据为基线,探讨bootstrap法在实际预警工作中的应用。结果:江苏省ILI流行曲线有2个高峰,冬春季高峰分布于11月—次年2月,夏季高峰分布于6—8月;bootstrap法、移动平均法和金标流行期一致性均较好;与移动平均法相比,bootstrap法对2009年初的流感流行相对更敏感,与金标流行期的一致性相对更高(kappa=0.728)。采用bootstrap法对2016年ILI%开展预警,与金标准结果接近。结论 :bootstrap法作为一种不依赖于ILI数据分布的计算密集型方法 ,可作为传统移动平均控制图预警方法的替代和补充,用于流感流行的监测和预警。
[Abstract]:Objective: to evaluate the value of bootstrap method in ILI early warning by using bootstrap method to estimate the incidence level of influenza-like illnesseses ILI (95% confidence interval 95 CI). Methods: the weekly surveillance data from 2005-2014 (except 2006, 2009) were used as the baseline. Using 30% of the highest positive rate of influenza virus in the past two years as the influenza epidemic standard, the gold standard of influenza epidemic period in 2006 and 2009 was determined. By retrospective analysis, the consistency of bootstrap method, moving average method and gold standard epidemic period was compared. Based on the weekly monitoring data from 2012 to 2015, this paper discusses the application of bootstrap method in practical early warning. Results: there are two peaks in ILI epidemic curve in Jiangsu Province, the peak in winter and spring is from November to February in the following year, and the peak in summer is in June to August with bootstrap method. Compared with the moving average method, the bootstrap method was more sensitive to the influenza epidemic in early 2009 and the consistency was higher than that of the gold standard epidemic period. The bootstrap method was used to warn the ILI% in 2016. Conclusion as a computationally intensive method independent of the distribution of ILI data, the bootstrap method can be used as a substitute and supplement of the traditional moving average control chart warning method for the surveillance and early warning of influenza epidemic.
【作者单位】： 江苏省疾病预防控制中心急性传染病防制所;
【基金】：国家卫生计生委科研基金(W201303) 重大新发传染病综合防控科技示范工程(BE2015714) 江苏省自然科学基金(BK20151595) 江苏省青年医学人才(QNRC2016539) 江苏省临床医学科技专项(BL2014081)
【分类号】：R511.7
，

本文编号：1536337