STATISTICAL METHODS FOR SURVIVAL DATA VIA FRAILTY MODELS

描述（摘自申请人的摘要）：这是一个R29重新- 研究者打算开发统计学的应用程序 分析多变量生存数据的方法。 脆弱的概念将是 用于对相关生存数据进行建模。 首先，方法将 在二元背景下开发，但大多数将扩展到 多元情境 本文提出了四种估计协同效应的方法。 变量效应，每个变量效应的预期表现均优于 现有的方法在效率方面。 大样本理论将是 将研究已开发的和小样本的统计特性。 此外，还提出了两种方法来估计 生存时间之间的联系。 这些结果将扩展到 研究脆弱模型和交叉设计的使用， 生存数据。 这项建议的另一个重要目标是， 这些模型在遗传流行病学中很有用。 申请人将 重点关注两个问题：（1）她提出了新的生存模式， 分析MZ和DZ双胞胎数据以获得可解释的参数 （2）遗传和环境影响;（3）申请人还 提出了新的多变量生存模型，可以处理不同的 家庭成员之间的相互关系。 她将扩展所提出的方法 在具体目标1中，对这些模型的参数进行推断 并将研究其统计特性;这是具体的 目标2. 来自NHLBI的心血管疾病数据集将用于 评估模型的适当性。 这些数据来自一项研究， 双胞胎 给定数据集脆弱性假设的评估方法 也将在本研究中进行评估。 这将通过 提出诊断图和拟合优度检验统计量，以及 作为研究上述方法的性能时， 违反了脆弱模型的假设。 三个数据集将 分析，以显示所提出的方法的使用，并产生一个更 完全了解使用的相对优点和局限性 在实际应用中的脆弱性模型。

DESCRIPTION (taken from the applicant's abstract): This is an R29 re- application in which the investigator intends to develop statistical methods to analyze multivariate survival data. Frailty concepts will be used to model the correlated survival data. Initially, methods will be developed in a bivariate context, but most will then be extended to the multivariate situation. Four methods are proposed for estimating co- variate effects each of which is expected to perform better than existing methods in terms of efficiency. Large sample theory will be developed and small sample statistical properties will be investigated. In addition, two methods have been proposed for estimating the association between survival times. These results will be extended to investigate the use of frailty models and crossover designs involving survival data. Another important objective of this proposal is to develop statistical models which are useful in genetic epidemiology. The applicant will focus on two issues: (1) she proposes new survival models for the analysis of MZ and DZ twin data to obtain interpretable parameters representing genetic and environmental effects; (2) the applicant also proposes new multivariate survival models that can handle different correlations among family members. She will extend the methods proposed in specific aim 1 to make inference on the parameters of these models and their statistical properties will be investigated; this is specific aim 2. A data set on cardiovascular disease from NHLBI will be used to assess the appropriateness of the models. These data are on a study of twins. The methods for assessing the frailty assumption for a given data set will also be evaluated in this study. This will be accomplished by proposing diagnostic plots and goodness-of-fit test statistics, as well as studying the performance of the preceding methods when the assumptions of frailty models are violated. Three data sets will be analyzed to show the use of the proposed methods and to yield a more complete understanding of the relative merits and limitations of the use of frailty models in practical applications.