Online Program Home
  My Program

Abstract Details

Activity Number: 160 - SPEED: Biometrics
Type: Contributed
Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
Sponsor: Biometrics Section
Abstract #324167 View Presentation
Title: A Semiparametric Latent Trait Model for Multiple Mixed Continuous, Binary and Ordinal Outcomes
Author(s): Sophie Yu-Pu Chen* and Alex Tsodikov
Companies: University of Michigan and University of Michigan
Keywords: Seimparametric latent variable model ; Mixed continuous and categorical outcomes ; latent trait model ; Factor analysis
Abstract:

This work addresses the challenge of latent trait measurement through multiple outcomes of mixed categorical and continuous types. Multiple outcomes (phenotype) are often collected when the construct of interest cannot be measured directly. A popular approach to latent variable models for mixed continuous, binary and ordinal outcomes assumes that the transformed outcomes follow a multivariate normal distribution. We propose a semiparametric shared latent variable model where a data-driven logistic link is used to accommodate continuous, ordinal, and binary outcomes. The model is used to provide a subject-specific measure of the latent trait, given the information observed on the subject. The proposed model avoids restrictive normality assumptions. The modeling framework is also generic with respect to the parametric distribution assumed for the trait. The proposed method is applied to measure pain centrality trait of patients undergoing hysterectomy as a treatment for pelvic pain and explain the heterogeneity of patients' reported outcomes. The method is compared with the ad-hoc 2011 Fibromyalgia (FM) Survey Criteria instrument designed to characterize a similar construct.


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

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association