深圳市公共场所集中空调冷却塔水嗜肺军团菌污染预警模型研究  

The study on the Early Warning Model Based on Bayesian Network for Legionella pneumophilia contamination in Cooling Water of Central Air Conditioning System from public places in Shenzhen City

作　　者：牟敬锋[1] 严宙宁[1] 孙健[1] 袁梦[1] 罗文亮[1] 刘可[1] 

机构地区：[1]深圳市南山区疾病预防控制中心,广东深圳518054

出　　处：《现代预防医学》2015年第10期1889-1892,1900,共5页Modern Preventive Medicine

基　　金：深圳市南山区科技创新局资助(2012060)

摘　　要：目的构建深圳市公共场所集中空调冷却塔水嗜肺军团菌污染预警模型,为预防控制军团菌病提供科学依据。方法于2013年随机选择深圳市50家安装有集中空调的公共场所进行冷却塔卫生学调查,并进行现场相关指标检测,同时采集50份冷却水水样进行微生物指标、理化指标和嗜肺军团菌检测。结合文献报道及专家咨询利用卫生学调查、现场检测和实验室检测结果构建贝叶斯网络预警模型。于2014年随机选择30家公共场所集中空调冷却塔,并收集相关数据,利用2014年收集数据对预警模型进行验证和评价。结果本研究构建的贝叶斯网络预警模型显示,对嗜肺军团菌最具有影响的因素有浊度、溶解性总固体、日照强度和防腐剂投放。该模型ROC曲线诊断临界点为0.338,利用该临界点进行嗜肺军团菌阳性诊断的灵敏度为100.0%,特异度为72.7%。结论本研究所得的贝叶斯预警模型预测的灵敏度和特异度符合风险预警的要求,可以快速判断出冷却水嗜肺军团菌污染情况。Objective This study aimed to establish the early warning model based on bayesian network for Legionella pneumophilia contamination in the cooling water of central air conditioning system from public places in Shenzhen city, so as tod provide a scientific basis for the prevention and control of Legionella disease. Methods 50 public places installed with central air conditioning were randomly selected to conduct hygienic investigation of cooling tower. Field test of various related indicators was conducted in Shenzhen City in 2013. 50 cooling water samples were collected for Legionella pneumophila detection at the same time. According to literature survey and consultation to experts, the early warning model based on bayesian network was constructed by the results of hygienic investigation, field test and laboratory test. 30 public places installed with central air conditioning were randomly selected to collect data in 2014. The data collected in this year was used to verify and evaluate the early warning model.Results The early warning model based on the bayesian network constructed in this study showed that the most important influencing factors of L. pneumophila contamination was turbidity, total dissolved solids, sunshine intensity and preservative. ROC curve of the model showed that the critical point was 0.338. The sensitivity of L. pneumophila diagnosis was 100%, and the specificity was 72.7% by using the critical point. Conclusion The sensitivity and specificity of the early warning model can meet the current requirements for early warning. This model can quickly detect L. pneumophila contamination in the cooling water.

关 键 词：冷却水 嗜肺军团菌 贝叶斯网络 预警 

分 类 号：R126.4[医药卫生—环境卫生学]