Statistical Issues in AIDS Research

描述：在发现艾滋病毒30年后，艾滋病的流行仍在继续，目前全世界估计有3400万人感染，每年新增250万人。预防和治疗艾滋病毒感染的研究变得越来越复杂，分析挑战也相应地变得复杂。这一应用旨在解决艾滋病毒/艾滋病研究中及时和重要的统计问题。特别是，将开发新的统计方法，用于(I)暴露前预防(PrEP)研究的数据分析，(Ii)采用复杂抽样和观察方案的研究的优化设计和分析，以及(Iii)合并研究的优化设计。首先，提出了解决PrEP研究中关键挑战的统计方法，即在不完全依从性的情况下推断药物疗效，结合不同的信息来源来估计依从性，以及总结依从性模式。这些方法包括因果建模、信息的贝叶斯综合和功能数据分析。拟议的方法有可能在一个被不同的试验结果阻碍的领域提供清晰度。其次，由于HIV研究经常使用昂贵的测试和分析方法，因此建议的研究将探索使用这种两阶段或多阶段抽样的研究的最佳抽样设计和有效的半参数和非参数方法。在区间删失数据的背景下，多阶段抽样和竞争风险问题也将被研究。这些方法将为利用昂贵化验的研究优化资源使用和最大限度地降低成本提供关键指导。最后，这项拟议的研究将研究用于分析汇集数据的统计方法，并将其具体应用于表位映射和在资源受限的情况下监测病毒学故障。前者是艾滋病毒疫苗开发的关键组成部分，后者有可能极大地改善发展中国家艾滋病毒感染者的保健。拟议的研究反映了调查人员在艾滋病毒/艾滋病研究领域的广泛和持续参与。所处理的统计问题是由艾滋病毒研究关键领域的应用推动的。概述的解决方案具有很高的创新性，直接适用于疫苗开发、暴露前预防试验以及其他艾滋病毒/艾滋病试验和研究方面的当前科学研究。

DESCRIPTION: Thirty years after identification of HIV, the AIDS epidemic continues with an estimated 34 million individuals currently infected worldwide and 2.5 million new infections each year. Research to prevent and treat HIV infection has grown increasingly sophisticated and the analytic challenges have become correspondingly complex. This application is intended to address timely and important statistical issues in HIV/AIDS research. In particular, novel statistical methods will be developed for (i) analysis of data from studies of pre-exposure prophylaxis (PrEP), (ii) optimal design and analysis of studies with complex sampling and observation schemes, and (iii) optimal design of pooling studies. First, statistical methods are proposed that address key challenges in PrEP studies, namely, inferring drug efficacy in the face of incomplete adherence, combining disparate sources of information to estimate adherence and summarizing adherence patterns. The approaches include causal modeling, Bayesian synthesis of information and functional data analysis. The proposed methods have the potential to provide clarity in a field which is hindered by disparate trial results. Second, becaue HIV studies often utilize expensive tests and assays, the proposed research will explore optimal sampling designs and efficient semiparametric and nonparametric methods for studies that use such two-phase or multi-phase sampling. The problems of multi-phase sampling and competing risks in the context of interval censored data will also be studied. These methods will provide critical guidance for optimizing the use of resources and minimizing costs in studies that utilize expensive assays. Finally, the proposed research will investigate statistical methods for the analysis of pooled data with specific applications to epitope mapping and monitoring for virologic failure in resource constrained settings. The former is a critical component of HIV vaccine development and the latter has the potential to greatly improve health care of HIV-infected individuals in developing countries. The proposed research reflects the extensive and ongoing involvement of the investigators in the field of HIV/AIDS research. The statistical problems addressed are motivated by applications in key areas of HIV research. The solutions outlined are highly innovative and directly applicable to current scientific research in vaccine development, trials of pre-exposure prophylaxis, and other HIV/AIDS trials and studies.