Advisor(s)
Charity Morgan
Committee Member(s)
Aneja Ritu
Justin Leach
Nengjun Yi
Xiaoxiao Zhou
Document Type
Dissertation
Date of Award
6-18-2026
Degree Name
Doctor of Philosophy (PhD)
School
School of Public Health
Department
Biostatistics
Abstract
This dissertation advances statistical methods for survival and spatial analysis to improve cancer outcomes research and address health disparities. The work is organized into two interrelated projects presented in three manuscripts. Project 1 focuses on progression-free survival at six months (PFS6) in Phase II oncology trials and is divided into two manuscripts. Manuscript 1 evaluates how different definitions of PFS6 (inclusion, exclusion, and Kaplan–Meier–based methods) influence early efficacy estimation under informative and non-informative censoring. Manuscript 2 extends this work by simulating informative and mixed censoring mechanisms using piecewise exponential models, quantifying the resulting bias in standard Kaplan–Meier estimates, and demonstrating the bias reduction achieved with inverse probability of censoring weighting (IPCW). Collectively, these two manuscripts form a coherent logic flow from defining and comparing PFS6 metrics to addressing bias correction under complex censoring. Project 2 (Manuscript 3) stands as an independent spatial analysis project that applies Bayesian disease-mapping methods to Alabama cancer registry data. It constructs census-tract–level maps of breast and lung cancer incidence using Poisson Besag–York–Mollié (BYM) models. To our knowledge, these are the first publicly available census-tract–level cancer incidence maps for Alabama. Prior state resources report rates only at the county level and for limited cancer sites. We compare direct age-adjusted incidence rates (AARs) with spatially smoothed Bayesian estimates and contrast two specifications: a baseline BYM model with tract-level covariates and an extended model with a neighborhood physician-access spillover term. Together, these studies provide practical guidance for handling informative censoring in survival endpoints and for producing robust small-area cancer estimates. The results strengthen the interpretability of PFS6 in Phase II trials and inform targeted cancer-control strategies at the census-tract level in Alabama.
Keywords
AARs;Besag–York–Mollié model;disease mapping;informative censoring;IPCW;PFS6
Recommended Citation
Wang, Lingling, "Statistical Innovations For Cancer Outcomes: Censoring In Progression-Free Survival And Bayesian Spatial Disease Mapping" (2026). ETDs from 2020-2029. 244.
https://digitalcommons.library.uab.edu/etd-2020s/244
