This dissertation explores Bayesian nonparametric (BNP) approaches for missing data imputation and causal inference with applications to electronic health records (EHR). The major contribution of the dissertation to the subject is 1) to propose a exible fully BNP model to impute missing covariates (confounders) in regression and comparative effectiveness analysis and further extend the approach to include auxiliary variables while keeping the imputation model always compatible with the inference model; 2) to propose BNP methods to estimate quantile causal effects by exibly modeling the propensity score and the distribution of potential outcomes. The methods developed are applied to a glucose study with missing covariates and a serum creatinine study comparing causal effects of two drug therapies using EHR data.
|Advisor:||Daniels, Michael J.|
|Commitee:||Ghosh, Malay, Presnell, Brett Douglas, Zhang, Lei|
|School:||University of Florida|
|School Location:||United States -- Florida|
|Source:||DAI-B 79/04(E), Dissertation Abstracts International|
|Keywords:||Bayesian additive regression trees, auxiliary variables, congenial malels, dirichlet process mixture motifs, multiple imputation, quantile causal effects|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be