Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 342 - Measurement Challenges and Innovations in Cognitive Aging and Dementia Research: Progress from the MEthods for LOngitudinal Studies of DEMentia (MELODEM) Initiative
Type: Topic-Contributed
Date/Time: Thursday, August 12, 2021 : 10:00 AM to 11:50 AM
Sponsor: Section on Statistics in Epidemiology
Abstract #317378
Title: Integrating Latent Variable Modeling, Statistical Learning, and Inverse Propensity Score Weighting to Identify Longitudinal Trajectory Phenotypes and the Underlying Disparity
Author(s): Chen-Pin Wang* and Mitzi M Gonzales and Helen Hazuda and Sudha Seshadri
Companies: University of Texas Health Science Center San Antonio and University of Texas Health Science Center San Antonio and University of Texas Health Science Center San Antonio and University of Texas Health Science Center San Antonio
Keywords: calibration; latent variable modeling; statistical learning; causal inference; disparity; dementia

Cognitive decline and decreasing gait speed are predictors of dementia and related disability and mortality. However, little is known about their joint trajectories to permit early identification of high-risk groups for timely intervention in older adults. This talk presents an integrated modeling framework to assess the phenotypes associated with joint trajectories of cognition and gait speed, their prediction for mortality, and determinants of ethnic disparities. Latent growth mixture modeling integrated with a targeted statistical learning technique is proposed to identify phenotypes associated with cognition and gait speed trajectories. Cross-validation metrics are used to evaluate trajectory phenotypes. Causal association between trajectory phenotype with mortality is verified by a statistical test of the latent structural model. In line with the Institute of Medicine’s definition of health disparity, we further incorporate inverse propensity score weights to infer ethnic disparity in the association between trajectory phenotypes with mortality and factors underlying the disparity. The proposed method is illustrated by data from the San Antonio Longitudinal Study of Aging.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2021 program