Statistical Problems in Multivariate Survival analysis

描述（由申请人提供）：本研究项目的广泛，长期目标是开发新的统计方法，用于分析心血管疾病研究和其他生物医学研究的多变量或相关故障时间数据。多变量失效时间数据在生物医学研究中自然出现。例如，在家庭研究中，记录了家庭成员的发病年龄;在流行病学队列研究中，对个别研究对象进行不同事件发生情况的跟踪;或在临床试验中，对患者进行重复复发事件的跟踪。在这些例子中，数据的一个共同特征是故障时间可能是相关的。有效的统计方法需要考虑这种相关性。这份竞争性续签申请中有5个具体目标。第一个目标是对来自病例队列研究的多变量失效时间数据进行统计推断。将考虑边际风险模型，并将开发用于参数估计的估计方程方法。第二个目的研究了一类多变量失效时间数据的半参数可加混合效应风险模型。提出了估计方程法和极大似然法。第三个目标是考虑时变系数率模型的复发事件数据。将考虑回归样条和惩罚回归样条来估计时变系数。第四个目标是关于多变量失效时间数据的具有灵活脆弱性分布的脆弱性模型。估计将基于筛似然估计程序。第五个目标是研究边际风险模型的推理过程与多变量失效时间数据的病例对照研究。将通过加权估计方程法进行参数估计。每种方法的优点和缺点将通过理论研究和模拟研究进行严格审查。将开发相关软件。将使用所提出的方法分析来自心血管疾病流行病学研究和其他生物医学研究的数据集。这项研究将为心血管疾病研究人员和其他生物医学研究人员提供有价值的新工具。这将有助于研究人员更好地了解与心血管疾病相关的危险因素。

DESCRIPTION (provided by applicant): The broad, long-term objective of this research project is to develop new statistical methodology for analyzing multivariate or correlated failure time data from cardiovascular disease research and other biomedical studies. Multivariate failure time data arise naturally in biomedical studies. For example, in family studies, the ages of disease occurrence are recorded for members of families; in epidemiological cohort studies, individual study subjects are followed for the occurrence of different events; or, in clinical trials, patients are followed for repeated recurrent events. A common feature of the data in these examples is that the failure times could be correlated. Valid statistical methods need to account for this correlation. There are 5 specific aims in this competing renewal application. The first aim concerns statistical inferences for multivariate failure time data from case-cohort studies. Marginal hazards model will be considered and an estimating equation approach will be developed for parameter estimation. The second aim studies a class of semi-parametric additive risk model with mixed effects for multivariate failure time data. An estimating equations approach and a maximum likelihood approach are proposed for estimation. The third aim considers time-varying coefficient rate models for recurrent event data. Regression splines and penalized regression splines will be considered to estimate the time-varying coefficient. The fourth aim concerns frailty models with flexible frailty distribution for multivariate failure time data. Estimation will be based on the sieve likelihood estimation procedure. The fifth aim investigates inference procedures for marginal hazards models with multivariate failure time data from case-control studies. Parameter estimation will be conducted through a weighted estimating equation approach. The strength and weakness of each proposed method will be critically examined via theoretical investigations and simulation studies. Related software will be developed. Data sets from epidemiologic studies on cardiovascular disease and other biomedical studies will be analyzed using the proposed methods. This research will provide valuable new tools to cardiovascular disease researchers and other biomedical researchers. It will help the researchers to better understand the risk factors associated with cardiovascular diseases.