|
Activity Number:
|
331
- Statistical and Practical Issues for Reproducible Molecular Prediction in Biomedical Studies
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2018 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
ENAR
|
|
Abstract #329104
|
Presentation
|
|
Title:
|
Estimating Directed Acyclic Graphs from High-Dimensional Data and Its Application in Biomarker Discoveries in Early Clinical Trials
|
|
Author(s):
|
Hua Zhong* and Jaehong Yu
|
|
Companies:
|
New York University and NYU School of Medicine
|
|
Keywords:
|
DAG;
biomarker
|
|
Abstract:
|
We develop a penalized likelihood estimation framework to estimate directed acyclic graphs (DAGs) from high-dimensional data. Our main contribution is a two-stage algorithm that can efficiently estimate sparse DAGs under the adaptive L1-penalized likelihood objective function with the acyclicity constraint. Simulations are presented to demonstrate the efficiency and flexibility of the proposed method. We then demonstrate a case study where we discover predictive biomarkers using DAG estimates from gene expression data in a Phase 1b trial to discriminate response to MDM2-antagonist therapy. The biomarkers are then assessed and validated in an independent MDM2-antagonist Phase 1b expansion trial. As a result, the gating assessment criteria of the discovered biomarkers are included in the recommendations to the interim data analysis and monitoring committee of the ongoing phase III trial. In addition, several practical issues of biomarker discovery and validation in early clinical trials will be discussed.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2018 program
|
