Date on Master's Thesis/Doctoral Dissertation
Bioinformatics and Biostatistics
Committee Co-Chair (if applicable)
rank-test; pseudo-value; jackknife; survival
This dissertation is composed of research projects that involve methods which can be broadly classified as either nonparametric or semiparametric. Chapter 1 provides an introduction of the problems addressed in these projects, a brief review of the related works that have done so far, and an outline of the methods developed in this dissertation. Chapter 2 describes in details the first project which aims at developing a rank-sum test for clustered data where an outcome from group in a cluster is associated with the number of observations belonging to that group in that cluster. Chapter 3 proposes the use of pseudo-value regression (Andersen, Klein, and Rosthøj, 2003) in combination with penalized and latent factor regression techniques for prediction of future state occupation in a multistate model based on high dimensional baseline covariates. Chapter 4 describes the development of an R package involving various rank based tests for clustered data which are useful in situations where the number of outcomes in a cluster or in a particular group within a cluster is informative. Chapter 5 explains the fouth project which aims at developing a covariate-adjusted rank-sum test for clustered data through alingned rank transformation.
Dutta, Sandipan, "Some contributions to nonparametric and semiparametric inference for clustered and multistate data." (2016). Electronic Theses and Dissertations. Paper 2454.