Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

High-dimensional Bias-corrected Inference, with Applications to fMRI Studies

Zhu, Xiaonan (2019) High-dimensional Bias-corrected Inference, with Applications to fMRI Studies. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Download (529kB) | Preview


In neuroimaging studies, measures of neural structure and function are used to try to predict clinical outcomes in adults. Identifying biomarkers that reflect underlying neuropathological processes can provide promising neural targets for future therapeutic interventions. This identification is typically done using linear or generalized linear models (GLM) with many covariates and relatively few subjects. Thus, regularization is used to select the salient covariates in the model. In this thesis, we compare the performance of the least absolute shrinkage and selection operator (LASSO) regression, adaptive LASSO regression, debiased LASSO regression, and regularized zero-inflated Poisson (ZIP) regression model in two simulation settings. The performance of LASSO regression with Poisson and Gaussian models are similar but for all these approaches the zero-inflated model outperforms the rest. We apply these approaches to the data from the Longitudinal Assessment of Manic Symptoms (LAMS) study. We then study the bias correction of GLM and the application on ZIP data. We apply a decorrelated score approach to address Poisson distributed data and introduce Cornish-Fisher correction to the decorrelated score test. In high-dimension settings, the Cornish-Fisher correction can improve the performance of decorrelated score test for ZIP data.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Zhu, Xiaonanxiz97@pitt.eduxiz97
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairIyengar,
Committee CoChairRen,
Committee MemberCheng,
Committee MemberHauskrecht,
Committee MemberBertocci,
Date: 25 September 2019
Date Type: Publication
Defense Date: 2 August 2019
Approval Date: 25 September 2019
Submission Date: 26 July 2019
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 65
Institution: University of Pittsburgh
Schools and Programs: Dietrich School of Arts and Sciences > Statistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: LAMS study, LASSO, ZIP, GLM, decorrelated score test, Cornish-Fisher
Date Deposited: 25 Sep 2019 16:08
Last Modified: 25 Sep 2019 16:08


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item