Statistical Methods for chronic Disease Research

描述（由申请人提供）：单个事件发生时间，例如，总生存时间是纵向癌症和HIV/AIDS研究中典型的主要终点。然而，仅此结果通常不足以捕获治疗（和/或其他协变量）可能产生的所有临床和经济影响。近年来，越来越多的人主张对治疗进行综合评估，这一努力带来了许多独特的统计挑战。这项研究的主要目标是解决慢性病研究中的一些挑战，并开发新的统计方法。主要工作将针对（1）不完整随访数据的成本和成本效益分析，以及（2）边际分析：多态过程中的事件间隔时间和复发事件。目前这两个领域的发展还不够，存在着巨大的知识差距。鉴于我们的医疗系统越来越受到有限资源的限制，成本评估目前已成为医学研究的重要组成部分，并已被纳入许多研究。这项工作将侧重于开发半参数估计和回归程序，以适应右删失数据。研究的第二个领域涉及由多种临床状态或复发事件代表的疾病过程的分析。现有的统计方法主要针对事件发生时间。这项研究将解决事件之间的时间，这在许多情况下是直接的科学兴趣。尽管这些长期存在的问题带来了挑战，但初步调查显示，优雅和实用的解决方案相当有希望。大样本特性的建议估计将严格调查。将进行广泛的模拟研究，以验证实际样本量下的这些建议。所提出的方法将被应用于一些临床研究。将开发便于使用的计算机程序，并提供给研究界。

DESCRIPTION (provided by applicant): A single time-to-event, e.g., overall survival time, has been the typical primary endpoint in longitudinal cancer and HIV/AIDS studies. However, this outcome alone is generally inadequate to capture all the impacts, clinical and economic, that a treatment (and/or other covariates) might have. Comprehensive assessment of treatments has been increasingly advocated in recent years, and this effort has posed many unique statistical challenges. The broad objective of this research is to address some of these challenges in chronic disease research and develop new statistical methods. Major efforts will be directed toward (1) cost and cost-effectiveness analyses with incomplete follow-up data and (2) marginal analysis'^ time-between- events in multi-state processes and recurrent events. Current developments in these two areas are inadequate and substantial gaps of knowledge exist. Given that our health care system is increasingly constrained with limited resources, nowadays cost evaluation is becoming an important component in medical research and has been integrated in many studies. This work will focus on developing semiparametric estimation and regression procedures that accommodate right-censored data. The second area of research concerns the analysis of disease processes represented by multiple clinical states or recurrent events. Existing statistical methods largely target at time-to-events. This research will address time-between-events, which are of direct scientific interest in many circumstances. Despite the challenges of these long-standing problems, preliminary investigations have shown considerable promise for elegant and practical solutions. Large-sample properties of the proposed estimators will be rigorously investigated. Extensive simulation studies will be performed to validate these proposals under practical sample sizes. The proposed methods will be applied to a number of clinical studies. User-friendly computer programs will be developed and made available to the research community.