Date of Completion
8-9-2020
Embargo Period
8-8-2025
Keywords
Longitudinal Data, Bayesian Joint Model, Alzheimer's Disease, SOMI, Transitional Model, High Dimensional Data, Human Microbiome Project
Major Advisor
Lynn Kuo
Associate Advisor
Ming-Hui Chen
Associate Advisor
Xiaojing Wang
Field of Study
Statistics
Degree
Doctor of Philosophy
Open Access
Open Access
Abstract
We propose three statistical methods for longitudinal ordinal outcomes and categorical outcomes. The proposed Bayesian joint model of longitudinal ordinal outcome and time-to-event data was applied to dementia data to evaluate the prediction accuracy of a staging model for memory impairment. The second method, the Bayesian stochastic model for multistage can be used to measure the correlation across different transition processes. This model was applied to the Human Microbiome Project (HMP) data to investigate the effect of baseline covariates on the change of microbiome type and evaluate the correlation of change in microbiome type and change in healthy status. The third method, the transitional model for high dimensional data is able to detect the significant covariates for multistage transitions. We applied this approach to the HMP data as well and we investigated the effect of microbiome genus data on the transition of health status.
Recommended Citation
Qi, Qi, "Statistical Methods for Longitudinal Data with Applications to Dementia and Human Microbiome Projects" (2020). Doctoral Dissertations. 2619.
https://digitalcommons.lib.uconn.edu/dissertations/2619