Combining theoretical and empirical network approaches to HIV transmission

描述（由申请人提供）：大量的实证研究表明，接触网络结构和行为风险对HIV和sti的传播有着深刻的影响。然而，从这些研究中收集到的各种见解尚未纳入一个总体框架，以了解性传播感染在异质人群中的地方性和流行病传播。有许多数学方法可以通过接触网络对疾病传播进行建模（例如，隔室模型；基于参与者的模型；网络模拟；指数随机图模型[也称为p*模型]），但大多数都是简化假设，与经验观察结果不一致。特别是，许多模型假定接触和行为模式在爆发期间保持不变。在这个探索性项目中，我们提出：(1)通过对8个实证网络研究数据的广泛统计分析，提出将人口结构和行为风险与STI传播动态联系起来的广泛假设；(2)使用强大的新数学模型来检验这些假设，该模型明确考虑了网络流动性的流行病学影响。我们的实证数据来自八个已完成的网络研究（六个在佐治亚州亚特兰大，一个在科罗拉多州的科罗拉多斯普林斯和亚利桑那州的弗拉格斯塔夫），以及一个在亚特兰大正在进行的研究，所有这些研究都使用相同的基本问卷。我们将汇集和合并这些研究，我们估计这些研究中80%的变量是一致的。我们将进行多变量分析，以确定哪些行为和人口因素对艾滋病毒/性传播感染的传播影响最大。然后，利用数学模型，结合从我们的八项研究中计算出的网络结构、行为风险和流行模式，我们将定量地探讨这些影响因素的含义，各种实证网络研究设计的功效，以及不同病原体在同一网络中的差异传播。公共卫生相关性：实地研究令人信服地表明，许多因素同时影响艾滋病毒/性传播感染，但这种直觉很难证明。通过一个共同的视角观察8个不同的社区，并采用最新一代的数学模型来反映观察到的模式，我们的目标是构建一个强有力地联系网络结构、行为风险和流行病学结果的理论框架。

DESCRIPTION (provided by applicant): Extensive empirical research has shown that the transmission of HIV and STIs is profoundly influenced by contact network structure and behavioral risks. Yet the diverse insights gleaned from these studies have not yet been integrated into a general framework for understanding the endemic and epidemic spread of STIs through heterogeneous populations. There are a number of mathematical approaches to modeling disease transmission through contact networks (for example, compartment models; actor-based models; network simulations; and exponential random graph models [also know as p* models]), but most make simplifying assumptions that are inconsistent with empirical observations. In particular, many models assume that contact and behavioral patterns remain constant for the duration of an outbreak. In this exploratory project, we propose (1) to develop broad hypotheses linking population structure and behavioral risk to STI transmission dynamics through extensive statistical analyses of data from eight empirical network studies, and (2) to test these hypotheses using powerful new mathematical models that explicitly consider the epidemiological impact of network fluidity. Our empirical data derive from eight completed network studies (six in Atlanta GA and one each in Colorado Springs CO and Flagstaff AZ), and one ongoing study in Atlanta, all of which used the same basic questionnaire. We will assemble and amalgamate these studies, and we estimate that 80% of the variables in each of these studies will be congruent. We will conduct multivariable analyses to determine which behavioral and population factors most strongly influence the transmission of HIV/STIs. Then, using mathematical models that incorporate the network structures, behavioral risks, and prevalence patterns calculated from our eight studies, we will quantitatively explore the implications of these influential factors, the efficacy of various empirical network study designs, and the differential transmission of diverse pathogens in the same network. PUBLIC HEALTH RELEVANCE: Field studies convincingly suggest that many factors simultaneously influence HIV/STI transmission, but such intuition is difficult to document. By viewing eight diverse communities through a common lens and adapting the latest generation of mathematical models to reflect the observed patterns, we aim to construct a theoretical framework that robustly relates network structure, behavioral risks, and epidemiological outcomes.