All ETDs from UAB

Advisory Committee Chair

Hemant K Tiwari

Advisory Committee Members

Inmaculada Aban

Daniel C Bullard

William M Geisler

Charles R Katholi

Nengjun Yi

Document Type


Date of Award


Degree Name by School

Doctor of Philosophy (PhD) School of Public Health


Chlamydia trachomatis (CT) infection is the most common sexually transmitted infec-tion. The burden is greatest in young, African American women. CT reinfections are common and increase risk for reproductive morbidity. Why some women are more sus-ceptible to reinfection is not well understood. Emerging evidence suggests immunologic and genetic variants may predict reinfection susceptibility. The primary goal of this dissertation is to evaluate immunologic and genetic vari-ants as predictors of CT reinfection in young, African American women using different modeling approaches, which has never been reported previously. To evaluate a much larger number of potential predictors than our sample size (a concept known as high di-mensionality), innovative statistical methods are needed to address the reduced statistical power. Furthermore, genetic data is inherently collinear. Even after using methods of linkage disequilibrium pruning, remaining variants may be correlated by chance due to high dimensionality. To address the potential limitation of high dimensionality in our evaluation of immunologic and genetic predictors of CT reinfection, we initially used a Bayesian hier-archical logistic regression model with weakly informative Cauchy priors. To address potential collinearity and incorporate gene-level biological structure, we next used a group spike-and-slab lasso model with mixture double-exponential priors. Finally, to in-corporate prior biological information, we used a two-stage pathway-structured predictive model using the spike-and-slab lasso with mixture double-exponential priors. We first showed that incorporating SNPs into a model of CT reinfection with a prori predictors improves prediction. We then demonstrated that incorporating gene structure using the group spike-and-slab lasso logistic regression model with mixture double exponential priors was a higher performing method for prediction than other Bayesian hierarchical penalized logistic regression methods. Finally, we illustrated that using Gene Ontology pathway-structured modeling and the spike-and-slab lasso model with mixture double exponential priors was highly predictive of CT reinfection. The findings demonstrate the utility of incorporating gene- and pathway-level structure to predictive models of CT reinfection and will help to advance CT prevention and control efforts.

Included in

Public Health Commons



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.