Generalized Meta-Analysis for Combining Disparate Risk Factor Information Across Studies: Inference on Multiple Regression Based Risk Prediction Models

• Activity Number: 551 - Risk Prediction Methods and Applications in Risk Stratified Prevention
• Type: Invited
• Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Prosenjit Kundu* and Runlong Tang and Nilanjan Chatterjee
• Companies: The Johns Hopkins University Bloomberg School of Public Health and The Johns Hopkins University Bloomberg School of Public Health and Johns Hopkins University
• Keywords: Data Integration; Generalized Method of Moments; Meta-Analysis; Missing Data; Semiparametric Inference; Risk Prediction

Abstract

One of the central components in the domain of risk prediction involves developing rich statistical models to predict the future risk of disease. The objective of this talk is to describe recent methods for inferring parameters associated with such models using disparate risk factor information available from multiple studies. Meta-analysis is a popular approach for synthesizing information on common parameters of interest across multiple studies due to logistical convenience and statistical efficiency. We will describe a generalization of this approach that allows estimation of parameters associated with a multiple regression model through meta-analysis of studies which may individually have information only on partial sets of the regressors. An application of the method will be illustrated through a real data example involving the development of a breast cancer risk prediction model using disparate risk factor information from multiple studies. Further, we will show how the proposed framework can be used for efficient analysis of data from two-phase epidemiologic designs.