|
Activity Number:
|
145
- Statistical Methods in Data Integration and Data Harmonization
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, July 31, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
ENAR
|
|
Abstract #322192
|
|
|
Title:
|
Generalized Meta-Analysis: Towards an Unified Paradigm for Model Building Through Integration of Disparate Data Sources
|
|
Author(s):
|
Nilanjan Chatterjee* and Runlong Tang and Prosenjit Kundu
|
|
Companies:
|
Johns Hopkins University and Johns Hopkins University and Johns Hopkins University
|
|
Keywords:
|
Case-control study ;
Semiparametric inference ;
Generalized method of moments ;
Missing Data ;
Risk prediction model ;
Precision Medicine
|
|
Abstract:
|
In the future, methods for synthesizing data from multiple disparate sources will be critical for developing robust disease prediction models with rich set of risk factor information. We propose to develop general methodology for conducting meta-analysis of estimates of parameters of multivariate risk models of potentially different dimensions. The proposed methodology will build on our recent study that provides general mathematical foundation for relating parameters of regression models in different dimensions. The framework allows identifying a set of constraints, or "moment" equations, for models in higher dimension based on information given on parameters of fitted reduced dimensional models. Identification of these constraints requires having information on joint distribution of all of the different factors in the underlying population. We propose generalized method of moment approach for inference and illustrate multiple applications.
|
Authors who are presenting talks have a * after their name.
