Analysis of binary long data in multi-center studies with missing covariates

 
   Lead Investigator:    Andrew Zhou
   Institution      :    University of Washington
   E-Mail           :    azhou@u.washington.edu
   Proposal ID      :    53


   Publications:


1  Chen B, Zhou A
   Marginal Method for Multi-level Incomplete Binary Data
   2010   Abstract  Joint Statistical Meeting 2010 , 

   ACCEPTED



2  Chen B, Zhou A
   Generalized Partially Linear Models for Incomplete Longitudinal Data In the Presence of Population-Level Information
   Journal Article  Biometrics,69:386-395,2013,PMC3715115

   PUBLISHED



3  Chen B , Yi G, Cook R,  Zhou XH
   Marginal Methods for Clustered Longitudinal Binary Data with Incomplete Covariates
   Journal Article  J of Stat Plan Inference,142:2819-2831,2012,PMC3690662

   PUBLISHED





   Proposal Description:
   Aim 1. Develop weighted GEEs to estimate the cross-sectional mean and  association parameters in longitudinal data in multi-center studies when some  subjects have missing covariate values. Aim 2: Apply the newly proposed methods  to the UDS database.    The overall goal of this research is to develop statistical methods for evaluating risk factors from multiple sources in estimating risk of progression to MCI and AD. The specific aims are:    Aim 1. To combine risk factors including demographics, family history of dementia, and neuropsychological test performance to improve prediction of subjects at risk of progression.     Aim 2. To develop a Markov regression model for transition between discrete clinical diagnosis states (normal cognition, MCI, AD) that allows for underlying variation in cognitive functioning via latent variables    Aim 3.  To use risk factors from multiple sources to identify disease subtype clusters within clinical diagnosis states including MCI and AD.