【作者】 于德宪；

【导师】 陈清；

【作者基本信息】 南方医科大学 ， 流行病与卫生统计学， 2008， 博士

【摘要】 研究背景登革热是由伊蚊传播的一种急性传染病，主要分布于热带和亚热带的国家和地区，是分布最广、发病人数最多的虫媒传染病之一，估计全球每年有1亿例感染者，对人群健康造成极大威胁。登革热在我国主要分布在东南沿海的广东、福建、海南和台湾四省，近年来在浙江等地也出现了局部流行。随着全球气候的不断变暖，登革热有向北逐渐扩散的趋势，同时，全球化趋势、自然生态系统的破坏也使得该传染病的预防和控制工作越发复杂。疾病监测是指长期、连续、系统地收集疾病的动态分布及其影响因素的资料，经过分析将信息上报和反馈，以便及时采取干预措施并评价其效果。疾病监测工作包括4个基本环节：收集资料、分析资料、反馈信息和利用信息。其中充分利用监测信息对疾病流行进行预测预警是疾病监测的主要目的之一。我国目前对登革热的监测包括人间疫情监测（疫情监测、血清学监测、病原学监测）和媒介监测（媒介密度监测、病毒监测）两方面。信息管理是疾病监测的核心环节。传统的监测信息管理无论是“逐级上报”还是“网络直报”模式都只能完成一般的属性信息管理工作，不能完成从地理概念出发的操作，往往缺乏有关疾病的地理空间信息，这就难以有效反映疾病的地理空间分布特征和空间关系，无法准确掌握疾病的流行规律，因此，尽快研究设计出方便实用的疾病监测信息管理系统成为迫在眉睫的事情。近年来地理信息系统技术发展迅速，其与公共卫生领域的结合不断紧密，它已成为处理、分析和可视化空间资料必不可少的工具。建立基于地理信息系统相关技术的信息管理系统是对传染病疫情适时监控、预警、快速反应的重要手段，它可以及时掌握传染病疫情的发展、变化，使疫情信息的传递更加及时、准确，使有关部门能够在第一时间内及时采取应急措施，减少不必要的损失。媒介监测对于登革热的预测具有重要意义。目前对媒介伊蚊的监测主要采用幼虫监测法，指标为房屋指数、容器指数和布雷图指数等；2005年5月提出的《全国病媒生物监测方案（试行）》将诱蚊诱卵器列为伊蚊的监测方法之一，但应用还不广泛。数据信息的管理是传染病监测过程中的重要一环。2004年之前，我国传染病疫情信息主要由医疗卫生机构使用纸质报告卡，通过邮寄或电话方式向区疾病预防控制机构报告，由其按全国统一的软件录入计算机的逐级上报的模式；从2004年元月开始，国家卫生部规定所有法定传染病和其他一些重要传染病疫情全部通过联网数据库“中国疾病预防控制信息系统”进行网络直报，但一些虫媒传染病如登革热等的媒介昆虫监测资料由各监测单位自行管理，并不统一上报。在预测预警方面，目前无论是国内还是国外，都还没有有效的方法可以对登革热流行进行预测预警。有相关研究认为，伊蚊房屋指数、容器指数和布雷图指数较高时，登革热流行的危险性增加，因此全球多数国家主要以这些指标作为登革热流行前的预警指标，我国要求平时将布雷图指数控制在20以下，而登革热流行时，则要求迅速将其控制在5以下。我国现行的伊蚊监测方法主要为幼虫监测法，这种方法最早是用于黄热病媒介埃及伊蚊的监测，存在工作量大，干扰因素多的缺点，也不完全适用于半野栖的白纹伊蚊。它主要是通过对房屋、庭院中积水容器里的伊蚊幼虫、蛹数量进行监测，进而对环境中的成蚊密度进行估算，但由于现代社会的发展，城市中的建筑主要以楼房为主，庭院和积水容器的数量越来越少，对它们的监测已不足以反映环境中的蚊虫密度。林立丰等设计的新型伊蚊监测工具诱蚊诱卵器已经作为常规监测方法之一在我国一些地区进行应用，这种装置对白纹伊蚊的特异性较强，但诱蚊诱卵效率并不高。而国外应用较广的伊蚊监测方法——诱卵杯法，也存在不能捕获成蚊和诱卵效率低的缺点。因此，改进传统的监测方法，提高蚊媒监测工具的诱蚊和诱卵效率，使之更准确的反映环境中的蚊虫密度和带毒情况，是对登革热进行有效预防和预测的先决条件。寻找合适的蚊虫诱引剂是解决方法之一。传统的登革热预警指标主要是蚊媒的监测数据，但仅用该指标预测效果不理想。例如，新加坡自1979年以来房屋指数控制在2以下和巴西布雷图指数控制在5以下，登革热仍不断流行。因为登革热是由媒介伊蚊传播的急性传染病，因此环境中伊蚊的密度以及影响伊蚊密度的一些因素如气候变化等都会对登革热的流行产生影响，若想对登革热进行较准确的预测预警，必须在登革热流行数据的基础上，结合气候因素、环境因素等建立多因素的预测预警模型。气候变化可以使蚊虫的地理分布范围发生变化，提高繁殖速度，增加叮咬率等而直接影响疾病传播。因此，监测气候因素的变化，进而分析气候因素在传染病发生中所起的作用，将有助于制定更有效的传染病预防控制策略。登革热主要由伊蚊进行传播，而蚊虫密度又受到气温、降雨量、湿度等气侯因素的影响，因此通过进行登革热与气侯因素的相关研究，可以建立基于气侯因素的登革热早期预测预警系统。鉴于上述背景和分析，我们进行了登革热监测方法的系列研究，旨在提高登革热监测的效能与预警水平，促进登革热预防与控制工作的科学发展。目的1．构建基于地理信息系统的登革热监测信息管理系统；2．筛选有效的诱引剂，提高白纹伊蚊的监测效果；3．分析气候因素对登革热流行的影响，为进一步建立登革热预警模型提供科学依据。方法1．基于GIS的登革热监测信息管理系统研究采用北京超图地理信息技术有限公司开发的SuperMap Objects5．2全组件式开发平台实现系统的构建。以Visual Basic6．0为集成环境，通过Active数据对象（ADO）、数据访问对象（DAO）和数据环境（DE）等软件工程技术利用SuperMapObjects核心组件实现登革热相关空间数据库的集成，通过调用动态函数库（DLL）实现GIS组件功能与数据库程序之问的数据传递和数据表现，并且构成统一的无缝界面。2．不同诱引剂对白纹伊蚊密度监测效果影响的研究2．1诱引剂的准备（1）容器颜色实验：对照组：诱蚊诱卵器中放入1张直径为10 cm的圆形滤纸，加入20ml过夜自来水，加盖，标记，备用；杯身黑色组：诱蚊诱卵器杯身上部2／3用黑色塑料包裹，其余同对照组。（2）浸出液实验：大黍组：称取3g新鲜大黍叶，剪碎，浸泡于盛有300ml过夜自来水的烧杯中，放在60℃水浴加热2h，然后用纱布过滤，所得液体即为大黍叶浸出液；轮胎组：废旧汽车轮胎块，剪成碎块，浸泡于一个有300ml过夜自来水的烧杯中，放在60℃水浴中加热2h，然后用纱布过滤，所得液体即为轮胎浸出液；对照组同（1）。（3）化学诱引剂实验：乳酸组：左旋乳酸与过夜自来水混合稀释为0．5mg/ml、1mg/ml、2mg／ml、5mg/ml、10mg/ml、20mg/ml溶液，备用；对照组同（1）。（4）酵母来源CO₂实验：实验一：分为实验组和对照组。实验组每个诱蚊诱卵器中加入50ml过夜自来水配制的25％蔗醣溶液和0．25g酵母粉，并在杯壁上部用双面胶贴一圈宽3cm的滤纸，加盖，标记，备用；对照组同（1）。实验二：实验分为五个组，分别为A组：自来水+酵母粉+蔗醣，B组：自来水+蔗醣，C组：自来水+酵母粉，D组：自来水，E：原诱蚊诱卵杯。其中ABCD组每个诱蚊诱卵杯中加50ml液体，杯壁上部用双面胶贴一圈宽3cm的滤纸。蔗醣浓度为25％，酵母粉为0．25g/杯。E组为对照组，同（1）。实验三：实验分为六个组，分别为A组：5％蔗糖溶液50ml+0．25g酵母粉，B组：15％蔗糖溶液50ml+0．25g酵母粉，C组：25％蔗糖溶液50ml+0．25g酵母粉，D组：35％蔗糖溶液50ml+0．25g酵母粉，E组：45％蔗糖溶液50ml+0．25g酵母粉，F组：对照组，同（1）。2．2布放和回收（1）室内实验：从野外获得白纹伊蚊蚊卵进行培养，羽化后3～4d，将小白鼠固定挂在蚊笼内供成蚊吸血，1d后取50只雌蚊放入蒙式蚊帐（1．2 m×1．8 m×1．5 m），在四个角落布放诱蚊诱卵器。连续布放5天后，观察计算诱到的蚊虫数，并将虫卵倒进白搪瓷盆，在放大镜下观察计算蚊卵数。每一实验重复2次。（2）室外实验：将准备好的诱蚊诱卵器布放到家属区周围的绿化带内和地下停车场里，每个布放点内诱蚊诱卵器间隔20cm，点与点间隔25m左右。在诱蚊诱卵器布放后的每天下午15～16点钟检查诱蚊诱卵情况，连续7天。检查时用10倍放大镜观察，记录捕获的蚊虫数及种类，完成后将其放回原来位置，7天后统一收回。3．气候因素对登革热的影响收集广州市历年蚊媒监测资料（人工小时捕蚊数）、登革热发病人数和气候资料，用SPSS13．0软件建立数据库及进行统计分析。蚊虫密度与气候因素的关系应用Spearson等级相关和逐步多元线性回归方法进行分析，登革热发病与白纹伊蚊密度、气候因素的关系应用主成分logistic回归方法进行分析。应用误差反向传播神经网络建立白纹伊蚊密度的预测模型。使用Matlab软件作为建模工具，以广州市1996至2000年间的气候和白纹伊蚊监测数据作为网络的学习训练样本，以2001年的数据作为预测样本。结果1．基于GIS的登革热监测信息管理系统研究本系统根据登革热数据管理的特点，简化了用户操作，使系统的易用性得到增强。主要的功能有以下几方面：（1）文件管理。（2）数据库建立（3）地图的创建与操作。（4）查询：图查属性，属性查图，图形和属性数据双向查询检索。（5）专题图的制作：根据用户选择的属性数据项进行专题图的显示。（6）空间分析：缓冲区分析。（7）进行有关属性数据的统计分析。（8）提供操作结果多种形式的输出。2．不同诱引剂对白纹伊蚊密度监测效果影响的研究（1）容器颜色实验：诱蚊诱卵器杯身变为黑色并没有增强其诱捕白纹伊蚊的能力，反而增加了蚊虫逃逸的概率，室内和室外实验中实验组蚊虫逃逸率分别为50％和66．7％，而对照组为0。（2）浸出液实验：大黍浸出液、轮胎浸出液在室内和室外实验中诱蚊阳性杯数、诱蚊数、诱卵数均与对照组相差不大，在诱蚊指数、诱蚊密度指数、诱卵指数、诱卵密度指数的比较上，差异均没有统计学意义（P>0．05）。（3）化学诱引剂实验：在室内乳酸实验中，随L-乳酸浓度的升高，诱蚊诱卵器捕获白纹伊蚊和蚊卵的数量均呈先升后降的趋势，在浓度为1mg/ml时诱引力最强，随后逐渐下降。乳酸各浓度组诱蚊指数、诱蚊密度指数、诱卵指数、诱卵密度指数与对照组比较差异均没有统计学意义（P>0．05）。室外现场实验的结果与实验室研究结果一致，1mg／ml L-乳酸的诱蚊指数、诱蚊密度指数、诱卵指数和诱卵密度指数分别为55．56％，1．8，55．56％，60．50，高于对照组的38．89％，1．14，50％，56．89，但差异均没有统计学意义（P>0．05）。（4）酵母来源CO₂实验实验一：诱蚊诱卵器随布放时间的延长，捕获白纹伊蚊的数量、诱蚊指数和诱蚊密度指数均呈逐渐上升的趋势。布放7天，CO₂组的诱蚊指数从40．74％增长到74．07％，诱蚊密度指数从1．64上升为3．10；对照组的诱蚊指数从18．52％增长到40．74％，诱蚊密度指数从1．20上升为1．36。CO₂组每天的诱蚊指数、诱蚊密度指数均高于对照组，其中2～7天的诱蚊指数和诱蚊密度指数两组比较差异有统计学意义（P<0．05）。实验二：实验组每天的诱蚊指数和诱蚊密度指数均为CO₂组最高，对照组较低。布放7天的诱蚊指数由高到低分别为CO₂组（89．66％）、蔗醣组（44．83％）、酵母组（40．00％）、自来水组（33．33％）和对照组（26．67％）。统计检验显示各组每天的诱蚊指数比较差异均有统计学意义（P<0．01）。布放7天的诱蚊密度指数由高到低分别为CO₂组（6．19）、蔗醣组（2．08）、酵母组（1．58）、自来水组（1．40）和对照组（1．75）。统计检验显示各组4～7天的诱蚊密度指数比较差异均有统计学意义（P<0．05）。实验三：布放7天的诱蚊指数由高到低分别为5％蔗醣浓度组（96．55％）、15％蔗醣浓度组（93．33％）、25％蔗醣浓度组（93．10％）、对照组（90．00％）、35％蔗醣浓度组（88．89％）、45％蔗醣浓度组（82．76％）。统计检验显示各组1～5天的诱蚊指数比较差异均有统计学意义（P<0．05）。布放7天的诱蚊密度指数由高到低分别为45％蔗醣浓度组（5．21）、5％蔗醣浓度组（5．04）、15％蔗糖浓度组（4．64）、35％蔗醣浓度组（4．08）、25％蔗醣浓度组（3．70）和对照组（2．52）。统计检验显示各组间每天的诱蚊密度指数比较差异没有统计学意义（P>0．05）。3．气候因素与登革热流行的关系（1）相关分析广州市在连续7年里人工小时捕蚊总数与同期的蒸发量、相对湿度、日照、温度、降雨量都存在正相关关系，而与气压有负相关关系（P<0．05）；而人工小时捕获白纹伊蚊数与同期的蒸发量、相对湿度、绝对湿度、日照、温度、降雨量存在正相关关系，与气压存在负相关关系（P<0．05）。（2）回归分析采用逐步多元回归的方法筛选影响蚊虫密度的主要气候因素，以人工小时捕蚊总数为因变量，以各气候因素为自变量，进行逐步多元回归分析，发现仅绝对湿度一个因素对蚊虫密度的影响有统计学意义，可以纳入回归方程。模型的复相关系数为0．835，决定系数为0．697。用方差分析对整个回归模型进行检验，F=215．839，P<0．001，说明模型具有统计学意义。经回归分析得到回归方程：（?）=362．230+0．835X以人工小时白纹伊蚊数为因变量，以各气候因素为自变量，进行逐步多元回归分析，结果显示日照（X₁）、风速（X₂）、绝对湿度（X₃）和温度（X₄）四个因素对白纹伊蚊密度的影响有统计学意义，可以纳入回归方程。模型的复相关系数为0．850，决定系数为0．723。用方差分析对整个回归模型进行检验，F=59．323，P<0．001，说明模型具有统计学意义。经回归分析得到方程：（?）=1．959+0．609X₁-0．186X₂-0．256X₃+0．349X₄（3）主成分logistic回归分析选中前4个主成分代替原来的自变量，特征根分别为3．652、2．490、1．150、0．905，累积贡献率为88．626％。通过主成分因子负荷矩阵可知，第一主成分主要包含原变量温度、日照、相对湿度、蒸发量、伊蚊密度、降雨量和绝对湿度的信息，第二主成分主要包含原变量绝对湿度、降雨量和伊蚊密度的信息，第三主成分主要包含原变量风速的信息，第四主成分主要包含原变量气压的信息。主成分logistic回归后，将标准自变量还原为原变量，得到应变量y与原自变量X₁-X₉的回归方程：（?）=exp（-3．3926-0．0005X₁+0．0241X₂-0．00004X₃-0．0001X₄+0．0253X₅+0．0341X₆+0．0165X₇-0．0012X₈+0．0202X₉）／1+exp（-3．3926-0．0005X₁+0．0241X₂-0．00004X₃-0．0001X₄+0．0253X₅+0．0341X₆+0．0165X₇-0．0012X₈+0．0202X₉）（X₁-X₉分别代表：气压、蒸发量、相对湿度、绝对湿度、日照、温度、风速、降雨量、捕蚊数）（4）基于人工神经网络技术建立白纹伊蚊密度预测模型建立的BP神经网络模型经25次学习和训练，误差从0．305539下降至2．93751×10^-14，预测符合率为80％。考虑到样本的容量比较小的原因，可以接受结果。结论基于GIS建立的登革热信息管理系统能够较好的将属性数据和空间信息结合，为相关数据的管理、展示和空间分析提供简单易用的操作平台。酵母来源的二氧化碳对白纹伊蚊有较强的吸引力，可以作为诱引剂与伊蚊监测工具结合应用，提高对蚊媒密度和带毒情况监测的准确度。气候因素中日照、风速、绝对湿度和温度对白纹伊蚊密度的影响较大，而温度、日照、蒸发量、白纹伊蚊密度与登革热的发生关系较为密切，有效的预警指标可以通过进一步研究从气候因素中进行提取；基于BP神经网络建立的白纹伊蚊密度预测模型有较好的预测效果，可以作为潜在的登革热预测预警模型进行更深入的研究。更多还原

【Abstract】 Dengue fever is an acute infectious disease transmitted by Aedes mosquitoes, which distributes primarily in tropical and subtropical countries and regions. In the world, it’s approximately one hundred million cases to be infected annually, and the disease has threatened human being health greatly. In our country, dengue fever mostly distributes in Guangdong, Fujian, Hainan and Taiwan provinces, while the disease also emerges in other provinces such as Zhejiang. With global climate getting warmer gradually, this disease tends to diffuse from the south to the north. Meanwhile, trend of globalization and destroy of natural ecosystem make the prevention and control of this epidemic more complicated. As Aedes mosquito is easily affected by many environmental factors, it’s difficult to use traditional measures to surveil and control the disease, so that the research and application of the new surveillance measures is urgent for the epidemiology staffs.Disease surveillance refers to collecting the information of the dynamic distribution and the influential factors consistently and systematically, and reporting the information after analysis in order to take preventive measures and evaluate the effects. Disease surveillance includes 4 basic steps, collecting, analyzing, reporting and using the information. Make full use of the information to predict the epidemics is the main purpose of disease surveillance. At present, the main surveillance methods of our country are larval survey, and the indices are house index, container index and Breteau index. The vector biology and control surveillance scheme includes mosq-ovitrap as a surveillance method, while its application is not extensive. The management of data information is an important link of the epidemic surveillance process. Before 2004, the epidemic disease information is mainly reported through paper reporting cards. After receiving the hospital’s letters or phone calls, the district disease prevention control center will type into the computer by mode of reporting to the next higher level of authority. From 2004, ministry of health regulates that all the legal epidemics and other important epidemic diseases must be reported directly through computer network data system "China disease prevention control information system". The insect-borne surveillance information of dengue fever is managed by individual surveillance unit. There is no effective means to pre-warning the dengue fever in the world. Some researches think that the danger increases when the house index, container index and Breteau index are high. So most countries refer to these indexes as the warning index. In our country, the Breteau index must be lower than 20 at ordinary times. When dengue fever breaks out, the index should be controlled under 5.The main problem is that with the development of the society, the traditional surveillance methods do not fit the current society environment, while the new effective method is not yet found out. The current surveillance method in our country is larval survey. The method is used to supervise the yellow fever vector Egyptian Aedes mosquito. Shortcoming of this method is too much work and interfering factors. This method supervises the number of Aedes mosquito larvae and chrysalides, and estimates the mosquito density of the environment. With the development of modern society, the buildings in the city are mostly storied buildings. Courtyards and containers are less and less. Supervising the courtyards and containers is not enough to reflect the mosquito density. The new mosa-ovitrap designed by Lin Lifen has become one of the regular surveillance methods in our country. This device is not good at trapping the larvae. Thus, improving the traditional surveillance method is the prerequisite of preventing and predicting the dengue fever. Proper mosquito attractant is one of the solving methods.Whether "reporting of grade by grade" or "reporting direct through the computer network" mode can just fulfill the ordinary attribute information management work, while cannot fulfill the work from geography concept. Lack of geography space information of the disease will lead to hardly reflecting geography space distribution characters and space relations, thus hardly controlling the epidemic rules of the disease. It’s urgent to design a convenient disease surveillance information management system. In recent years, GIS technology developed rapidly, and has become a necessary facility to deal with and analyze the space information. Establishing an information management system based on GIS is an important instrument of surveillance, pre-warning and rapid-reacting to the epidemics. It can handle the development and change of epidemic in time, transfer the information exactly, so that the relevant agency can adopt emergency measures in time, reducing the unnecessary loss.During the course of urbanization, traditional dengue fever early-warning indexes do not suit modern society environment. Singapore has controlled the house index under 2 and the Brazil Breteau index under 5 Since 1979, while the dengue fever still exists. As the dengue fever is an acute infectious disease transmitted by Aedes mosquitoes, the density of the Aedes mosquito and some factors influencing the density will affect the prevalence of the dengue fever. We must establish a early forecasting and warning system based on dengue fever data, climate factors, environment factors, and so on. Among these factors, climate factors are important to dengue fever, and it can affect the transmission of insect-borne infectious disease. Thus, supervising changes of the climate factors and analyzing the function of these factors will help to formulate more effective strategy to control infectious diseases. Dengue fever is an insect-borne infectious disease and transmitted by Aedes mostly. Mosquito density is affected by some climate factors such as temperature, rainfall and humidity. Research on the relation between dengue fever and climate factors will help to establish a climate-based early-warning system.At present, main problems include that surveillance cannot reflect the mosquito density exactly, people cannot analyze the disease combining with local environment factors such as vegetation and climate, and cannot do the research on the space relation of the surveillance data. So it’s hard to catch the dynamic change rules. In order to provide an easy-handling platform for data management, demonstration and space analysis, we have developed a series of research on dengue fever surveillance methods. In the meantime, we research the attractant for Aedes albopictus and the relation between dengue fever and climate factors. These researches offer theory base for the improvement of dengue fever surveillance system.1. Objective（1） Using geographic information system platform software to develop an information management system for dengue fever.（2） Screening potential mosquito attractant, and enhancing the surveillance effect of Aedes albopictus.（3） Analyzing the influence of climate factor, providing scientific proof for establishing dengue fever early warning mode.2. Methods2.1 Study of dengue fever information management systemWe use SuperMap Objects5.2 platform software to develop this system, with Visual Basic6.0 software as integration environment. As developing this system, the integration of dengue fever space data base is finished through SuperMap Objects core component and some software engineering technique such as active data object, data access object and data environment. After that, using dynamic link library to implement data transmission and data display between GIS component function and data base program, then constituted unified seamless interface.2.2 Influence of different attractants to Aedes albopictus surveillance effectiveness2.2.1 Preparation of attractants（1） Experiments of container colorControl group: In the bottom of Mosq-ovitraps, put a sheet of 10cm filter paper and infunde 20ml overnight tap water, then head up and mark. Black bottle group: Mosq-ovitraps are wrapped up by a plastic bag in black at its upper 2/3 part.（2） Experiments of infusionsPanicum maximum infusion group: 3g fresh Panicum maximum leaves, shorn to broken bits and soak to a beaker which filled 300ml overnight tap water, then the beaker is put to water bath of 60 degree and warmed up for two hours, after that, the liquid is filtered by gauze, which is Panicum maximum infusion. Tyre leachate group: used automobile tyre is shorn to broken bits and soaked to a beaker which filled 300ml overnight tap water, then the beaker is put to water bath of 60 degree and warmed up for two hours, after that, the liquid is filtered by gauze, that is tyre infusion. Control group was same to （1）.（3） Experiments of chemical attractantsL-lactic acid group: L-lactic acid is diluted to 0.5mg/ml, 1mg/ml, 2mg/ml, 5mg/ml, 10mg/ml and 20mg/ml through overnight tap water. Control group was same to （1）.（4） Experiments of yeast generated CO₂ Experiment one: Add the mixture that consist of 50ml overnight tap water, 25%cane sugar solution and 0.25g yeast powder to each Mosq-ovitrap in experiment group, then stick a wrap of 3cm width filter paper in the upper part of Mosq-ovitrap. Comparison group was same to （1）.Experiment two: This test had five groups, they were A group, B group, C group, D group and E group, and A group adds overnight tap water, yeast powder and cane sugar to each Mosq-ovitrap, B group adds overnight tap water and cane sugar, C group adds overnight tap water and yeast powder, D group adds overnight tap water, E group is control group which is same to （1）.Experiment three: This test has six groups. A group adds 50ml 5% cane sugar solution and 0.25g yeast powder to each Mosq-ovitrap, B group adds 50ml 15% cane sugar solution and 0.25g yeast powder, C group adds 50ml 25% cane sugar solution and 0.25g yeast powder, D group adds 50ml 35% cane sugar solution and 0.25g yeast powder, E group adds 50ml 45% cane sugar solution and 0.25g yeast powder, and F group is control group which is same to （1）.2.2.2 Installation and retrieval Mosq-ovitraps（1） Indoor experimentsAfter being fed mouse blood for 2 days, 50 adult female mosquitoes are placed into a large mosquito net（l.2×1.8×1.5m³） for testing. Mosq-ovitraps with different attractants are installed at four corners of the mosquito net for 5days. Then count the number of the trapped adult mosquitoes and collect the eggs laid in a white porcelain bowl and estimated with a 10x magnifier. The experiment is done for 2 times.（2） Outdoor experimentsMosq-ovitraps with different attractants are set at intervals of 20cm, and two close spots are at intervals of about 20m. Count the number of adult mosquitoes and eggs every afternoon for 7 days. Count and record mosquitoes captured and the egg positively trapped every day.2.3 The influence of climate factor to dengue feverWe collect mosquito monitoring data, dengue fever incidence and climate data for several years in Guangzhou, then use SPSS 13.0 statistic software to establish data base and do statistic analysis. The relation between mosquitoes density and climate factors is studied through statistic means of Spearson rank correlation and multiple linear regression. The relation between dengue fever and Aedes albopictus density, climate factors is analyzed by principal component logistic analysis.Use Matlab software to establish a model of BP neural network to forecast the density of Aedes albopictus. Training sample is data from 1996 to 2000, and predictive sample is data in 2001.3. Results3.1 Development of dengue fever monitoring information management system basing on GISAccording to the feature of public health, we simplify user operation and function of this system, make it more easy to apply. Main functions are as follows:（1） Document management: open workspace, close workspace, save workspace, save another workspace, print, exit, and so on.（2） Establishing data base: the system can establish new database, as well as change old database.（3） Drawing and operation to map.（4） Inquest: through map to inquire about attribute, through attribute to inquire about map.（5） Manufacture of special subject map: According to special attribute data to form subject map.（6） Space analysis: buffer area analysis. （7） Some statistic analysis to attribute data such as maximum value, minimum value, total sum and standard deviation, and so on.（8） The system can output different format results.3.2 Influence of different attractants to Aedes albopictus density monitoring effectiveness（1） Experiments of container colorIn the experiments, we find that making containers black could not strengthen their ability to catch more Aedes albopictus. Adversely, it raises the number of escaped mosquitoes. The escaping rates of mosquitoes in indoor and outdoor experiments are 50% and 66.7% respectively, while the number in Comparison group was 0.（2） Experiments of infusionThere is no significant difference in mosquito-trap positive index, mosquito-trap density index, oviposition index and oviposition density index between Panicum maximum infusion, tyre infusion and control group（P>0.05）.（3） Experiments of chemical attractantsIn lactic acid experiments, with lactic acid concentration heightening, the number of Mosq-ovitraps-caught mosquitoes and eggs shows a trend of increasing to decreasing. At the concentration of 1mg/ml, the ability of lactic acid in attracting mosquito is most powerful. But in both indoor and outdoor experiments, there is no significantt difference in mosquito-trap positive index, mosquito-trap density index, oviposition index and oviposition density index between different concentration of lactic acid and control group（P>0.05）.（4） Yeast generated CO₂ experimentsExperiment one:By prolonging installation time, mosquito-trap positive index and mosquito-trap density index in two groups show a trend of increasing by degrees, but variation amplitude in experiment is above control group. In the period of 7 days, mosquito-trap positive index in experiment group varies from 40.74% to 74.07%, mosquito-trap density index varies from 1.64 to 3.10; mosquito-trap positive index in control group varies from 18.52% to 40.74%, mosquito-trap density index varies from 1.20 to 1.36. After mosq-ovitraps are installed for 2-7 days, there is significant difference in mosquito-trap positive index and mosquito-trap density index between experiment group and control group（P<0.05）.Experiment two:By prolonging the installation time, mosquito-trap positive index and mosquito-trap density index in all groups show a trend of increasing by degrees. From high to low, mosquito-trap positive indexes are 89.66%（CO₂ group）, 44.83%（cane sugar group）, 40.00%（yeast group）, 33.33%（tap water group）, 26.67%（control group）; and mosquito-trap density indexes are 6.19（CO₂ group）, 2.08（cane sugar group）, 1.58（yeast group）, 1.40（tap water group）, 1.75（control group）. There is significant difference in mosquito-trap positive indexes among all groups each day, and mosquito-trap density index has statistical significance in all groups for 4-7 days （P<0.05）.Experiment three:By prolonging the installation time, mosquito-trap positive index and mosquito-trap density index in all groups show a trend of increasing by degrees. From high to low, mosquito-trap positive indexes are 96.55%（5% sugar group）, 93.33%（15% sugar group）, 93.10% （25 sugar group）, 90.00%（control group）, 88.89%（35% sugar group）, 82.76%（45% sugar group）; and mosquito-trap density indexes are 5.21（45% sugar group）, 5.04（5% sugar group）, 4.64（15% sugar group）, 4.08（35% sugar group）, 3.70（25% sugar group）, 2.52（control group）. There is significant difference in mosquito-trap positive index among all groups for 1-5 day, and mosquito-trap density index has no statistical significance in all groups for each day （P>0.05）.3.3 The influence of climate factor to dengue fever（1） Correlation analysisThere is positive correlation relation between the number of captured mosquito and homeochronous evaporation capacity, relative humidity, absolute humidity, sunshine, precipitation （P<0.05）, but there is negative correlation relation with pressure（P<0.05）.（2） Regression analysisWe use stepwise regression to screen main climate factors of affecting mosquito density, and the result shows that only absolute humidity could enter regression equation. Correlation coefficient of the regression model is 0.835, and coefficient ofdetermination is 0.697. The final equation is: Y =362.230 + 0.835X.We regarded Aedes albopictus density as dependent variable and regarded climate factors as independent variable to run stepwise regression analysis. The result shows that sunshine（x₁）, wind speed（x₂）, absolute humidity（x₃） and temperature（x₄） could enter regression equation. Correlation coefficient of the regression model is0.850, and coefficient of determination is 0.723. The final equation is: （?） =1.959 + 0.609X₁ -0.186X₂-0.256X₃ + 0. 349X₄.（3） Principal component logistic regression analysis.We choose fore-four principal constituent to take the place of independent variable, and their characteristic roots are 3.652、2.490、1.150、0.905 respectively, accumulation contribution rate is 88.626%. After principal component logistic regression analysis, we make principal component independent variable convert to incipient independent variable, and got the regression equation: （?）=exp（-3.3926-0.0005X₁+0.0241X₂-0.00004X₃-0.0001X₄+0.0253X₅+0.0341X₆+0.0165X₇-0.0012X₈+0.0202X₉） / 1+exp（-3.3926-0.0005X₁+0.0241X₂-0.00004X₃-0.0001X₄+0.0253X₅+0.0341X₆+0.0165X₇-0.0012X₈+0.0202X₉）（X₁-X₉: pressure, evaporation capacity, relative humidity, absolute humidity, sunshine, temperature, wind speed, precipitation, number of captured Aedes albopictus ）（4） Using ANN method to establish a forecasting model for Aedes albopictus densityFor 25 times learning and training, network error decreases from 0.305539 to 2.93751×10^-14, and predictive coincidence rate is 80%.4. ConclusionDengue fever information management system may be a operation platform that apply for data management, data display and data space analysis and so on. Meanwhile, it also apply for the related analysis of other infectious disease. So, it can become a general GIS platform system for public health research.Comparing with the control group, Panicum maximum infusion, tyre infusion could not attract Aedes albopictus entering and oviposition in mosq-ovitrap. To a certain extent, L-lactic acid could raise the ability of trap in captured Aedes albopictus, but there is no significant difference between experiment and control group, so its ability in attracting Aedes albopictus need to be confirmed in the further study. Yeast generated CO₂ had strong attractive to Aedes albopictus.There is positive correlation relation between the number of captured Aedes albopictus and homeochronous evaporation capacity, relative humidity, absolute humidity, sunshine, precipitation （P<0.05）, but there is negative correlation relation with pressure（P<0.05）. Sunshine, wind speed, absolute humidity and temperature could significantly influence the density of Aedes albopictus in environment （P<0.05）. Dengue fever has close relation with some climate factors such as temperature, sunshine, evaporation capacity and density of Aedes albopictus. The ANN forecasting model displays a good ability in forecasting Aedes albopictus density. So, we should study it further and establish dengue fever early warning model based the ANN method.更多还原