Directory Calendar News Careers Alumni Giving

Defenses

Upcoming Dissertation Defenses

Biostatistics Dissertation Defense
Joanne Beer - Predicting Clinical Variables from Neuroimages Using Fused Sparse Group Lasso Biostatistics Dissertation Defense
Joanne Beer - Predicting Clinical Variables from Neuroimages Using Fused Sparse Group Lasso
Tue 7/17/2018 1:00PM - 3:00PM
Public Health 7139, Peterson Seminar Room

Joanne Beer of the Department of Biostatistics defends her dissertation on "Predicting Clinical Variables from Neuroimages Using Fused Sparse Group Lasso".


Tue 7/17/2018
Biostatistics Dissertation Defense
Christopher Keener - Power and Sample Size Determination for Stepped Wedge Cluster Randomized Trials Biostatistics Dissertation Defense
Christopher Keener - Power and Sample Size Determination for Stepped Wedge Cluster Randomized Trials
Wed 7/18/2018 10:00AM - 12:00PM
Public Health 7139, Peterson Seminar Room

Christopher Keener of the Department of Biostatistics defends his dissertation on "Power and Sample Size Determination for Stepped Wedge Cluster Randomized Trials".


Wed 7/18/2018

Maximize Recent Master's Defenses

Minimize Recent Dissertation Defenses

Biostatistics Dissertation Defense

Song Zhang: Diagnostic Accuracy Analysis for Ordinal Competing Risks Outcomes Using ROC Surface

Wednesday 12/6 12:00PM - 2:00PM
Public Health 7139, Peterson Seminar Room

Song Zhang of the Department of Biostatistics defends her dissertation on "Diagnostic Accuracy Analysis for Ordinal Competing Risks Outcomes Using ROC Surface".



Graduate faculty of the University and all other interested parties are invited to attend


ABSTRACT:

Many medical conditions are marked by a sequence of events or statuses that are associated with continuous changes in some biomarkers.  However, few works have evaluated the overall accuracy of a biomarker in separating various competing events. Existing methods usually focus on a single cause and compare it with the event-free controls at each time.  In our study, we extend the concept of ROC surface and the associated volume under the ROC surface (VUS) from multi-category outcomes to ordinal competing risks outcomes. We propose two methods to estimate the VUS. One views VUS as a numerical metric of correct classification probabilities representing the distributions of the diagnostic marker given the subjects who have experienced different cause-specific events. The other measures the concordance between the marker and the sequential competing outcomes.  Since data are often subject to lost of follow up, inverse probability of censoring weight is introduced to handle the missing  disease status due to independent censoring.  Asymptotic results are derived using counting process techniques and U-statistics theory.  Practical performances of the proposed estimators in finite samples are evaluated through simulation studies and the procedure of the methods are illustrated in two real data examples.

Public Health Significance: ROC curve has long been treated as a gold standard in assessing the accuracy of continuous predictors to binary outcomes.  Our proposed methods extend its utilization to multi-category events outcomes in the presence of competing risks censoring, as well as independent censoring. Our methods aim to assess a global accuracy of a biomarker's predictive power to each of events simultaneously, especially, to which stages of disease progression that patients would land by a specific time in follow-up. 

 

Last Updated On Monday, April 09, 2018 by Valenti, Renee Nerozzi
Created On Monday, October 23, 2017

© 2018 by University of Pittsburgh Graduate School of Public Health

Login  |  Sitemap