Semiparametric Models, Methodologies and Related Theory for Analysis of Censored Survival Data

本计画主要探讨半参数回归模型、方法与理论。研究者和他的合作者的目的是开发半参数有效估计的比例风险混合治愈模型的回归参数。该方法被扩展到另一类半参数治愈模型。他们提出了一个一般类的混合变换模型，这是常见的医学和计量经济学文献，并研究他们通过估计方程以及一些非参数平滑技术。此外，他们计划开发基于边际线性变换模型的多变量失效时间数据的估计方程，并提出一类具有协变量调整的多变量失效时间数据的独立性检验。他们还从病例队列设计中推导出了相对简单的生存数据分析方法，并讨论了通过投影方法进行更有效的参数估计。 这里研究的统计问题的动机是在生物医学科学，工程科学，社会学，经济学和遗传学的应用。该项目开发适当的统计模型，推理方法和数学理论。研究结果可用于临床试验和流行病学研究的设计，特别是在癌症，心血管疾病的研究中，分析工程可靠性和市场渗透率数据，评估由于不可测量的影响，如家族遗传效应，调整一些环境因素后，故障时间之间的关联。

This project addresses issues related to semi-parametric regression models, methods and theories with censored survival data. The investigator and his collaborators aim to develop semi-parametric efficient estimators of the regression parameters for the proportional hazards mixture cure model. The approach is extended to another class of semi-parametric cure models. They propose a general class of mixture transformation models, which are common in medical and econometrics literature, and study them via estimating equations as well as some nonparametric smoothing techniques. In addition, they plan to develop estimating equations for multivariate failure time data based on marginal linear transformation models and propose a class of independence tests for multivariate failure time data with the adjustment of covariates. They also derive relatively simple method for analysis of survival data from case-cohort design and discuss more efficient parameter estimation via the projection method. The statistical problems studied here are motivated by applications in biomedical sciences, engineering sciences, sociology, economics and genetics. The project develops appropriate statistical models, inferential methods and mathematical theory. The results can be used to facilitate design of clinical trials and epidemiological studies, particularly in studies of cancer, cardiovascular diseases, to analyze engineering reliability and market penetration data, to assess the association among failure times due to unmeasured effects, such as familial genetic effects, after adjusting some environmental factors.