Statistical Methods in Current Cancer Research

描述(由申请人提供)：这项研究的广泛、长期目标是发展统计方法，用于分析纵向癌症研究中的审查失败次数和不完全重复测量。这一竞争性更新应用的具体目标包括：(1)研究基于半参数回归模型的半参数回归模型的有效的似然方法；(2)构建用于不完全重复测量的回归分析的基于秩次的推理程序；(3)开发基于数量性状的截尾年龄表型或不完全重复测量的遗传连锁检测的统计方法，并在连锁分析中评估全基因组的统计意义；(4)探索在截尾失效时间表型的关联分析中适应群体分层的策略。其中一些是长期存在的统计学问题，而另一些则是当前癌症研究中新出现的问题。拟议的解决方案建立在合理的统计原则基础上。利用计数过程鞅理论、现代经验过程理论和半参数有效性理论，对新估计和检验统计量的渐近性质进行了严格的研究。通过计算机模拟，将广泛评估它们在实际环境中的运行特性。建议的推理程序的用处将通过真实的医学研究来说明，其中一些研究是在北卡罗来纳大学进行的。将开发实施新方法的软件，供公众使用。这项研究不仅将推动纵向数据分析、生存分析和统计遗传学领域的发展，而且将为癌症研究人员提供有价值的新工具。

DESCRIPTION (provided by applicant): The broad, long-term objectives of this research are the developments of statistical methods for the analysis of censored failure times and incomplete repeated measures from longitudinal cancer studies. The specific aims of this competing renewal application include: (1) investigation of efficient likelihood-based methods for semi-parametric regression models for possibly correlated failure times subject to right or interval censoring; (2) construction of rank-based inference procedures for the regression analysis of incomplete repeated measures; (3) development of statistical methods for detecting genetic linkage based on censored age-of onset phenotypes or incomplete repeated measures of a quantitative trait and for assessing genome-wide statistical significance in linkage analysis; (4) exploration of strategies for accommodating population stratification in the association analysis of censored failure time phenotypes. Some of these topics are longstanding statistical problems while others are emerging issues from current cancer studies. The proposed solutions are built on sound statistical principles. The asymptotic properties of the new estimators and test statistics will be studied rigorously with the use of counting-process martingale theory, modern empirical process theory and semi-parametric efficiency theory. Their operating characteristics in practical settings will be evaluated extensively through computer simulation. The usefulness of the proposed inference procedures will be illustrated with real medical studies, some of which are carried out at the University of North Carolina. The software implementing the new methodologies will be developed for public use. This research will not only advance the fields of longitudinal data analysis, survival analysis and statistical genetics, but will also provide valuable new tools to cancer researchers.