Activity Number:
|
700
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biopharmaceutical Section
|
Abstract #316223
|
View Presentation
|
Title:
|
Patient-Specific Meta-Analysis with Application to a Genomic Prostate Cancer Diagnostic
|
Author(s):
|
Michael Crager* and Nan Zhang and Tara Maddala
|
Companies:
|
Genomic Health, Inc. and Genomic Health, Inc. and Genomic Health, Inc.
|
Keywords:
|
meta-analysis ;
prediction ;
cancer diagnostics
|
Abstract:
|
Patient-specific meta-analysis (PSMA, Crager and Tang, J. Appl. Stat. 2014) assesses prognosis or prediction of a clinical event for an individual patient using multiple studies. The method makes predictions for a patient by combining prediction-related estimates for the patient from each study, with weighting based on estimate precision. Fixed effects PSMA combines estimates from studies using inverse variance weighting. Random effects PSMA treats between-study differences as random. Both methods produce point estimates and confidence intervals. Simulation studies suggest the random effects model gives valid confidence intervals with 6 or more studies. The method can be used with proportional hazards regression, logistic regression, and linear models of continuous numeric outcome. An example application combines information from 2 validation studies of a prostate cancer diagnostic for assessing the probability of adverse pathology using a 17-gene genomic assay of biopsy tissue. PSMA estimates are more precise and reflect the body of evidence collected across these studies, resulting in better discrimination between aggressive and indolent disease.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2015 program
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.