Analyse multidimensionnelle de durées de vie

拟议研究计划的总体目标是多变量故障时间的联合分析。医学实验和后续研究提供了多变量失败时间的几种结构。这种结构的例子包括依赖截断的数据、连续事件之间的间隔时间和双变量当前状态数据。通常，这些模式中的每一种都需要特定的推断程序，随着新的多变量生存数据结构的出现，需要新的统计推断程序。这项研究计划的目的是开发分析多变量故障时间的新结构的方法。处理的问题可分为四类： -关联度量的非参数估计，如Kendall‘s tau， -生存函数的估计， -经常性事件均值函数的估计和 -开发针对Copulas等半参数模型的拟合优度检验。 将对拟议方法的理论、实践和计算方面进行调查。

The overall objective of the proposed research program is the joint analysis of multivariate failure times. Medical experiments and follow up studies provide several structures of multivariate failure times. Examples of such structures include data subject to dependent truncation, gap times between successive events and bivariate current status data. Typically, each of these pattern requires specific inference procedures and as new multivariate survival data structures arise, new statistical inference procedures are needed. The aim of this research program is to develop methodology to analyse new structures of multivariate failure times. The treated problems can be divided into four categories: - the nonparametric estimation of association measures such as Kendall's tau, - the estimation of survival functions, - the estimation of recurrent events mean function and - the development of goodness-of-fit tests for semi-parametric models such as copulas. The theoretical, practical and computational aspects of the proposed methodology will be investigated.