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

Longitudinal Multivariate Normative Comparisons and Accuracy Improvement Metrics for Competing Endpoints

Wang, Zheng (2020) Longitudinal Multivariate Normative Comparisons and Accuracy Improvement Metrics for Competing Endpoints. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Download (2MB) | Preview


Motivated by the Multicenter AIDS Cohort Study (MACS), we are interested in developing methods to address several challenges in analyzing the data. The objectives of the research include developing dementia classification procedures from longitudinal data to control family-wise error, modeling covariates effects on time to dementia, and evaluating the improvement of diagnostic accuracy when including a new biomarker in the model.

In order to properly define an event of interest, we adapt the cross-sectional multivariate normative comparison (MNC) method, which controls family-wise error by accounting for the inter-correlations among all covariates, to a longitudinal setting. Longitudinal MNC is constructed based on multivariate mixed effects models when hypothesis testing happens by the conclusion of study.

In practice, patients may require visit-by-visit classification. Prompt feedback can guide treatments for longer survival. We propose to modify longitudinal MNC statistics to build visit-by-visit test statistics. Then, based on predicted number of visits from survival model and Poisson regression, we can apply Bonferroni-type correction to control family-wise error for this prospective classification scenario.

Lastly, we examine a new biomarker's contribution to diagnostic accuracy for competing-risk outcomes. The net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) were originally proposed to characterize accuracy improvement in predicting a binary outcome, when new biomarkers are added to regression models. These two indices have been extended from dichotomous outcomes to multi-categorical and survival outcomes. We extend the NRI and IDI to competing-risk outcomes, by adopting the definitions of the two indices for multi-category outcomes. The ``missing'' category due to independent censoring is handled through the inverse probability weighting.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Wang, Zhengzhw61@pitt.eduzhw610000-0002-7201-0430
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairCheng,
Committee MemberIyengar,
Committee MemberChen,
Committee MemberChung-Chou,
Date: 8 June 2020
Date Type: Publication
Defense Date: 26 March 2020
Approval Date: 8 June 2020
Submission Date: 26 March 2020
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 112
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: Competing risks; Cumulative incidence function; Family-wise error rate; Integrated discrimination improvement; Multivariate mixed-effect model; Net reclassification improvement
Date Deposited: 08 Jun 2020 17:09
Last Modified: 08 Jun 2020 17:09


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item