A Generalized Model for Post-transplant Survival

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant) The applicant's long-term career goal is to become an independent investigator in analytical methodologies for clinical and health services research. She will dedicate her career to conducting quantitative studies and promoting the proper use of analytical methodologies in these research areas. This career development award will provide her with initial support for achieving her career goals. There are three goals for her career development over the next five years. First, she will develop a general survival regression model that is more encompassing than any of the semi-parametric regression models currently used in the analysis of liver transplantation data. The proposed research will result in less restrictive assumptions and provide a systematic way of constructing survival models in liver transplantation studies. The products of the research include a new model that can be used to provide estimates on post-transplant survival, which in turn can be used to estimate the optimal timing for liver transplantation. Her second objective is to develop expertise in clinical applications of survival analysis for patient-oriented research. The third goal is to become an educator to clinical researchers in analytical methodologies. While many other questions are also important in the area of transplantation (such as which organ allocation method is optimal and what is the appropriate time to list a patient for transplant), many of the clinically related questions require accurate and clinically relevant predictions of survival. The proposed research will focus on: Aim 1: Development of a general survival regression model. Aim 2: Development of diagnostic procedures for the general survival regression model. Aim 3: Assessment of the impact of different survival models on the prognosis in liver transplantation. The proposed general survival model that reduces to both the additive and multiplicative model as special cases. Moreover, the covariate effect on hazard in this model can be either constant or time-varying. Estimation of the model parameters will through the penalized log-partial likelihood with double penalty terms. Inference on the effect of any given covariate will also be assessed.