基于控制图法的流感暴发流行预警模型探讨  

Exploration on Precaution Model of Influenza Outbreak and Prevalence Based on Control Chart

作　　者：黄娇[1,2] 王高帅[1] 杨召[1] 张红艳[1] 耿梦晗[1] 吴丽娟[1] 杜瑜[1] 王重建[1] 

机构地区：[1]郑州大学公共卫生学院,郑州450001 [2]华中科技大学同济医学院公共卫生学院,武汉430030

出　　处：《医学与社会》2013年第9期26-28,共3页Medicine and Society

基　　金：国家自然科学基金;编号为81001293;河南省教育厅自然科学基金;编号为2010B330004

摘　　要：目的:探讨控制图法在流感暴发流行预警中的应用。方法:利用某省2006-2011年的流感疫情监测数据建立基于控制图法的预警模型,通过受试者工作特征(ROC)曲线筛选最佳预警界值;采用2012年流感疫情数据验证模型的预警判别效能。结果:该省2006-2012年流感发病率随时间变化呈上升趋势(P<0.01)。流感暴发流行的最佳预警界值为P85,ROC曲线下面积为0.958(95%CI:0.863-0.993),灵敏度为100.00%,特异度为91.67%。在P85的预警界值下,该模型的灵敏度为97.22%,特异度为75.00%、阳性预测值为89.74%、阴性预测值为92.31%,ROC曲线下面积为0.861(95%CI:0.737-0.941)。结论:在P85的预警界值下,基于控制图法的模型具有较好的预警判别效能。Objective： To explore the potential application of control chart to predict the outbreak and prevalence of influenza. Methods： According to the theory of control chart, we established the precaution model using the epidemic data of influenza in a province from January I st, 2006 to December 31 st, 2011, and the optimization alert threshold was screened by the receiver operating characteristic （ROC） curve. Then the predictive performance was tested and verified by analyzing the data in 2012. Results： The incidence of influenza showed a tendency of increase with the change of time （ P 〈 0. O1 ）. The optimization alert threshold of control chart was Ps5, and the sensitivity, specificity, and the area under the ROC curve were 100.00%, 91.67%, and 0. 958, respectively. Further analysis showed that the sensitivity, specificity, positive predictive value, and negative predictive value, and the area under of ROC curve were 97.22% , 75.00% , 89.74% , 92.31% , and 0. 861 respectively. Conclusions ： When the optimization alert threshold was P85, the precaution model of influenza outbreak and prevalence shows a highly predictive performance based on the control chart theory.

关 键 词：流感 预警 控制图 模型 

分 类 号：R181.2[医药卫生—流行病学]