Model selection in multivariate semiparametric regression

Zhuokai Li, Hai Liu, Wanzhu Tu

Research output: Contribution to journalArticle

Abstract

Variable selection in semiparametric mixed models for longitudinal data remains a challenge, especially in the presence of multiple correlated outcomes. In this paper, we propose a model selection procedure that simultaneously selects fixed and random effects using a maximum penalized likelihood method with the adaptive least absolute shrinkage and selection operator penalty. Through random effects selection, we determine the correlation structure among multiple outcomes and therefore address whether a joint model is necessary. Additionally, we include a bivariate nonparametric component, as approximated by tensor product splines, to accommodate the joint nonlinear effects of two independent variables. We use an adaptive group least absolute shrinkage and selection operator to determine whether the bivariate nonparametric component can be reduced to additive components. To implement the selection and estimation method, we develop a two-stage expectation-maximization procedure. The operating characteristics of the proposed method are assessed through simulation studies. Finally, the method is illustrated in a clinical study of blood pressure development in children.

Original languageEnglish (US)
JournalStatistical Methods in Medical Research
DOIs
StateAccepted/In press - Jan 1 2017

Fingerprint

Semiparametric Regression
Multivariate Regression
Model Selection
Shrinkage
Random Effects
Tensor Product Splines
Multiple Outcomes
Joint Model
Expectation Maximization
Fixed Effects
Likelihood Methods
Operating Characteristics
Semiparametric Model
Correlation Structure
Blood Pressure
Mixed Model
Selection Procedures
Nonlinear Effects
Longitudinal Data
Variable Selection

Keywords

  • adaptive group least absolute shrinkage and selection operator
  • adaptive least absolute shrinkage and selection operator
  • expectation-maximization algorithm
  • mixed effects
  • multivariate data

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability
  • Health Information Management

Cite this

Model selection in multivariate semiparametric regression. / Li, Zhuokai; Liu, Hai; Tu, Wanzhu.

In: Statistical Methods in Medical Research, 01.01.2017.

Research output: Contribution to journalArticle

@article{f06f3ed5298249eeb2a51bff5e23a1b5,
title = "Model selection in multivariate semiparametric regression",
abstract = "Variable selection in semiparametric mixed models for longitudinal data remains a challenge, especially in the presence of multiple correlated outcomes. In this paper, we propose a model selection procedure that simultaneously selects fixed and random effects using a maximum penalized likelihood method with the adaptive least absolute shrinkage and selection operator penalty. Through random effects selection, we determine the correlation structure among multiple outcomes and therefore address whether a joint model is necessary. Additionally, we include a bivariate nonparametric component, as approximated by tensor product splines, to accommodate the joint nonlinear effects of two independent variables. We use an adaptive group least absolute shrinkage and selection operator to determine whether the bivariate nonparametric component can be reduced to additive components. To implement the selection and estimation method, we develop a two-stage expectation-maximization procedure. The operating characteristics of the proposed method are assessed through simulation studies. Finally, the method is illustrated in a clinical study of blood pressure development in children.",
keywords = "adaptive group least absolute shrinkage and selection operator, adaptive least absolute shrinkage and selection operator, expectation-maximization algorithm, mixed effects, multivariate data",
author = "Zhuokai Li and Hai Liu and Wanzhu Tu",
year = "2017",
month = "1",
day = "1",
doi = "10.1177/0962280217690769",
language = "English (US)",
journal = "Statistical Methods in Medical Research",
issn = "0962-2802",
publisher = "SAGE Publications Ltd",

}

TY - JOUR

T1 - Model selection in multivariate semiparametric regression

AU - Li, Zhuokai

AU - Liu, Hai

AU - Tu, Wanzhu

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Variable selection in semiparametric mixed models for longitudinal data remains a challenge, especially in the presence of multiple correlated outcomes. In this paper, we propose a model selection procedure that simultaneously selects fixed and random effects using a maximum penalized likelihood method with the adaptive least absolute shrinkage and selection operator penalty. Through random effects selection, we determine the correlation structure among multiple outcomes and therefore address whether a joint model is necessary. Additionally, we include a bivariate nonparametric component, as approximated by tensor product splines, to accommodate the joint nonlinear effects of two independent variables. We use an adaptive group least absolute shrinkage and selection operator to determine whether the bivariate nonparametric component can be reduced to additive components. To implement the selection and estimation method, we develop a two-stage expectation-maximization procedure. The operating characteristics of the proposed method are assessed through simulation studies. Finally, the method is illustrated in a clinical study of blood pressure development in children.

AB - Variable selection in semiparametric mixed models for longitudinal data remains a challenge, especially in the presence of multiple correlated outcomes. In this paper, we propose a model selection procedure that simultaneously selects fixed and random effects using a maximum penalized likelihood method with the adaptive least absolute shrinkage and selection operator penalty. Through random effects selection, we determine the correlation structure among multiple outcomes and therefore address whether a joint model is necessary. Additionally, we include a bivariate nonparametric component, as approximated by tensor product splines, to accommodate the joint nonlinear effects of two independent variables. We use an adaptive group least absolute shrinkage and selection operator to determine whether the bivariate nonparametric component can be reduced to additive components. To implement the selection and estimation method, we develop a two-stage expectation-maximization procedure. The operating characteristics of the proposed method are assessed through simulation studies. Finally, the method is illustrated in a clinical study of blood pressure development in children.

KW - adaptive group least absolute shrinkage and selection operator

KW - adaptive least absolute shrinkage and selection operator

KW - expectation-maximization algorithm

KW - mixed effects

KW - multivariate data

UR - http://www.scopus.com/inward/record.url?scp=85043694032&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85043694032&partnerID=8YFLogxK

U2 - 10.1177/0962280217690769

DO - 10.1177/0962280217690769

M3 - Article

JO - Statistical Methods in Medical Research

JF - Statistical Methods in Medical Research

SN - 0962-2802

ER -