Electronic Theses and Dissertations

Identifier

6640

Date

2020

Document Type

Thesis

Degree Name

Master of Science

Major

Biostatistics

Committee Chair

Yu Jiang

Committee Member

Meredith Ray

Committee Member

Hongmei Zhang

Abstract

Mediation analyses study the effect of exposures on outcomes thorugh mediators. With the develoment of high throughput technologies, high-dimenisonal data are increasinbl produced. To better control fro both Type I and II errors in high dimensional data, a screening method, emscreening, was developed using cross-validation. Through simulation studies, we examined the sensitivity and specificity of emscreening in different scenarios and compared them to two commonly used methods: the false-discovery rate (FDR) and Bonferroni methods. The propsed method showed higher sensitivity and comparable specificyt with the number of mediators tested was large (5,000 or 500,000). Overall, emscreening can identify significant mediators without losing statistical power. To illustrate the use of emscreening, we applied the emscreening method to screen the potential mediation effects of DNA methylation in the association between maternal smoking during pregnancy and early childhood asthma in the offsrping in the Isle of Wight (IoW) multi-generation cohort.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Download

Share

COinS