李翔,戚立松,戴明,俞智勇,杨学明
北仑出入境检验检疫局穿山办事处,浙江 宁波 315012
摘要:目的 探讨船舶外来医学媒介生物传入的风险评估方法。方法通过建立医学媒介风险logistic回归模型,同时采用层次聚类分析以实例数据进行医学媒介风险预测,与医学媒介实际检出情况进行比对,以受试者工作特征曲线(ROC曲线)比较2种方法的预测效果。结果船员国籍、本地季节和历史医学媒介疫情是外来医学媒介生物传入的危险因素。医学媒介风险logistic回归模型的预测灵敏度为50%,特异度为77%;层次聚类分析法灵敏度为11%,特异度为93%;以ROC曲线比较2种方法的预测效果,其ROC曲线下面积AUC分别为0.64和0.52,Logistic回归模型预测效果略高于层次聚类分析。结论2种预测方法(AUC)均在0.7以下,准确性不高,为国际航行船舶外来媒介生物风险评估及预警系统的建立提供一种思路。
关键词:外来病媒;国际航行船舶;层次聚类分析
中图分类号:R181.2;R195.1 文献标识码:B
Analysis of risk factors associated withintroduction of exotic
medical vectors by international navigationships
LI Xiang, QI Li-song, DAI Ming, YUZhi-yong,YANG Xue-ming
Ningbo Beilun Entry-exit Inspection andQuarantine Bureau, Ningbo, Zhejiang 315012, China
Abstract: Objective To explore a new method of risk assessment associated with theintroduction of exotic medical vectors by international navigationships. Methods The logistic regressionmodel and hierarchical cluster analysis were applied to conduct arisk assessment of introduction of exotic medical vectors byinternational navigation ships. The forecast significance betweenthe logistic regression model and hierarchical cluster analysis bythe receiver operator characteristic curve ROC wasevaluated. Results Risk factors included:crew nationality, vector-positive records and arriving season.Thesensitivity and specificity index of the logistic regression modelwas 50% and 77%.They were 11% and 93% by hierarchical clusteranalysis.The area under the ROC curve (AUC) of the logisticregression model and hierarchical cluster analysis was 0.64 and0.52. The former was slightly higher than the latter. Conclusion The area under the ROC curve (AUC) of thelogistic regression model and hierarchical cluster analysis is bothunder 0.7,which have lower statistically significant,but provides atrain of thought for the construction of risk assessment and earlywarning system.
Key words: Exotic medicalvectors;International navigation ship;Hierarchical ClusterAnalysis
《中国国境卫生检疫杂志》2012年8月刊