Premium
Latent variable models with nonparametric interaction effects of latent variables
Author(s) -
Song Xinyuan,
Lu Zhaohua,
Feng Xiangnan
Publication year - 2013
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.6065
Subject(s) - latent variable , latent variable model , bivariate analysis , latent class model , nonparametric statistics , univariate , econometrics , markov chain monte carlo , structural equation modeling , statistics , outcome (game theory) , bayesian probability , medicine , computer science , mathematics , multivariate statistics , mathematical economics
Renal disease is one of the common complications of diabetes, especially for Asian populations. Moreover, cardiovascular and renal diseases share common risk factors. This paper proposes a latent variable model with nonparametric interaction effects of latent variables for a study based on the Hong Kong Diabetes Registry, which was established in 1995 as part of a continuous quality improvement program at the Prince of Wales Hospital in Hong Kong. Renal outcome (outcome latent variable) is regressed in terms of cardiac function and diabetes (explanatory latent variables) through an additive structural equation formulated using a series of unspecified univariate and bivariate smooth functions. The Bayesian P‐splines approach, along with a Markov chain Monte Carlo algorithm, is proposed to estimate smooth functions, unknown parameters, and latent variables in the model. The performance of the developed methodology is demonstrated via a simulation study. The effect of the nonparametric interaction of cardiac function and diabetes on renal outcome is investigated using the proposed methodology. Copyright © 2013 John Wiley & Sons, Ltd.