Penalized count data regression with application to hospital stay after pediatric cardiac surgery

Zhu Wang, Shuangge Ma, Michael Zappitelli, Chirag Parikh, Ching Yun Wang, Prasad Devarajan

Research output: Contribution to journalArticle

Abstract

Pediatric cardiac surgery may lead to poor outcomes such as acute kidney injury (AKI) and prolonged hospital length of stay (LOS). Plasma and urine biomarkers may help with early identification and prediction of these adverse clinical outcomes. In a recent multi-center study, 311 children undergoing cardiac surgery were enrolled to evaluate multiple biomarkers for diagnosis and prognosis of AKI and other clinical outcomes. LOS is often analyzed as count data, thus Poisson regression and negative binomial (NB) regression are common choices for developing predictive models. With many correlated prognostic factors and biomarkers, variable selection is an important step. The present paper proposes new variable selection methods for Poisson and NB regression. We evaluated regularized regression through penalized likelihood function. We first extend the elastic net (Enet) Poisson to two penalized Poisson regression: Mnet, a combination of minimax concave and ridge penalties; and Snet, a combination of smoothly clipped absolute deviation (SCAD) and ridge penalties. Furthermore, we extend the above methods to the penalized NB regression. For the Enet, Mnet, and Snet penalties (EMSnet), we develop a unified algorithm to estimate the parameters and conduct variable selection simultaneously. Simulation studies show that the proposed methods have advantages with highly correlated predictors, against some of the competing methods. Applying the proposed methods to the aforementioned data, it is discovered that early postoperative urine biomarkers including NGAL, IL18, and KIM-1 independently predict LOS, after adjusting for risk and biomarker variables.

Original languageEnglish (US)
Pages (from-to)2685-2703
Number of pages19
JournalStatistical Methods in Medical Research
Volume25
Issue number6
DOIs
StatePublished - Dec 1 2016

Fingerprint

Pediatrics
Count Data
Biomarkers
Cardiac
Surgery
Thoracic Surgery
Length of Stay
Regression
Negative Binomial
Variable Selection
Elastic Net
Penalty
Poisson Regression
Kidney
Ridge
Acute Kidney Injury
Acute
Siméon Denis Poisson
Urine
Penalized Regression

Keywords

  • Enet
  • Mnet
  • negative binomial regression
  • Poisson regression
  • Snet
  • variable selection

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability
  • Health Information Management

Cite this

Penalized count data regression with application to hospital stay after pediatric cardiac surgery. / Wang, Zhu; Ma, Shuangge; Zappitelli, Michael; Parikh, Chirag; Wang, Ching Yun; Devarajan, Prasad.

In: Statistical Methods in Medical Research, Vol. 25, No. 6, 01.12.2016, p. 2685-2703.

Research output: Contribution to journalArticle

Wang, Zhu ; Ma, Shuangge ; Zappitelli, Michael ; Parikh, Chirag ; Wang, Ching Yun ; Devarajan, Prasad. / Penalized count data regression with application to hospital stay after pediatric cardiac surgery. In: Statistical Methods in Medical Research. 2016 ; Vol. 25, No. 6. pp. 2685-2703.
@article{2ea444164ce34a08a3adf43ea456c7bc,
title = "Penalized count data regression with application to hospital stay after pediatric cardiac surgery",
abstract = "Pediatric cardiac surgery may lead to poor outcomes such as acute kidney injury (AKI) and prolonged hospital length of stay (LOS). Plasma and urine biomarkers may help with early identification and prediction of these adverse clinical outcomes. In a recent multi-center study, 311 children undergoing cardiac surgery were enrolled to evaluate multiple biomarkers for diagnosis and prognosis of AKI and other clinical outcomes. LOS is often analyzed as count data, thus Poisson regression and negative binomial (NB) regression are common choices for developing predictive models. With many correlated prognostic factors and biomarkers, variable selection is an important step. The present paper proposes new variable selection methods for Poisson and NB regression. We evaluated regularized regression through penalized likelihood function. We first extend the elastic net (Enet) Poisson to two penalized Poisson regression: Mnet, a combination of minimax concave and ridge penalties; and Snet, a combination of smoothly clipped absolute deviation (SCAD) and ridge penalties. Furthermore, we extend the above methods to the penalized NB regression. For the Enet, Mnet, and Snet penalties (EMSnet), we develop a unified algorithm to estimate the parameters and conduct variable selection simultaneously. Simulation studies show that the proposed methods have advantages with highly correlated predictors, against some of the competing methods. Applying the proposed methods to the aforementioned data, it is discovered that early postoperative urine biomarkers including NGAL, IL18, and KIM-1 independently predict LOS, after adjusting for risk and biomarker variables.",
keywords = "Enet, Mnet, negative binomial regression, Poisson regression, Snet, variable selection",
author = "Zhu Wang and Shuangge Ma and Michael Zappitelli and Chirag Parikh and Wang, {Ching Yun} and Prasad Devarajan",
year = "2016",
month = "12",
day = "1",
doi = "10.1177/0962280214530608",
language = "English (US)",
volume = "25",
pages = "2685--2703",
journal = "Statistical Methods in Medical Research",
issn = "0962-2802",
publisher = "SAGE Publications Ltd",
number = "6",

}

TY - JOUR

T1 - Penalized count data regression with application to hospital stay after pediatric cardiac surgery

AU - Wang, Zhu

AU - Ma, Shuangge

AU - Zappitelli, Michael

AU - Parikh, Chirag

AU - Wang, Ching Yun

AU - Devarajan, Prasad

PY - 2016/12/1

Y1 - 2016/12/1

N2 - Pediatric cardiac surgery may lead to poor outcomes such as acute kidney injury (AKI) and prolonged hospital length of stay (LOS). Plasma and urine biomarkers may help with early identification and prediction of these adverse clinical outcomes. In a recent multi-center study, 311 children undergoing cardiac surgery were enrolled to evaluate multiple biomarkers for diagnosis and prognosis of AKI and other clinical outcomes. LOS is often analyzed as count data, thus Poisson regression and negative binomial (NB) regression are common choices for developing predictive models. With many correlated prognostic factors and biomarkers, variable selection is an important step. The present paper proposes new variable selection methods for Poisson and NB regression. We evaluated regularized regression through penalized likelihood function. We first extend the elastic net (Enet) Poisson to two penalized Poisson regression: Mnet, a combination of minimax concave and ridge penalties; and Snet, a combination of smoothly clipped absolute deviation (SCAD) and ridge penalties. Furthermore, we extend the above methods to the penalized NB regression. For the Enet, Mnet, and Snet penalties (EMSnet), we develop a unified algorithm to estimate the parameters and conduct variable selection simultaneously. Simulation studies show that the proposed methods have advantages with highly correlated predictors, against some of the competing methods. Applying the proposed methods to the aforementioned data, it is discovered that early postoperative urine biomarkers including NGAL, IL18, and KIM-1 independently predict LOS, after adjusting for risk and biomarker variables.

AB - Pediatric cardiac surgery may lead to poor outcomes such as acute kidney injury (AKI) and prolonged hospital length of stay (LOS). Plasma and urine biomarkers may help with early identification and prediction of these adverse clinical outcomes. In a recent multi-center study, 311 children undergoing cardiac surgery were enrolled to evaluate multiple biomarkers for diagnosis and prognosis of AKI and other clinical outcomes. LOS is often analyzed as count data, thus Poisson regression and negative binomial (NB) regression are common choices for developing predictive models. With many correlated prognostic factors and biomarkers, variable selection is an important step. The present paper proposes new variable selection methods for Poisson and NB regression. We evaluated regularized regression through penalized likelihood function. We first extend the elastic net (Enet) Poisson to two penalized Poisson regression: Mnet, a combination of minimax concave and ridge penalties; and Snet, a combination of smoothly clipped absolute deviation (SCAD) and ridge penalties. Furthermore, we extend the above methods to the penalized NB regression. For the Enet, Mnet, and Snet penalties (EMSnet), we develop a unified algorithm to estimate the parameters and conduct variable selection simultaneously. Simulation studies show that the proposed methods have advantages with highly correlated predictors, against some of the competing methods. Applying the proposed methods to the aforementioned data, it is discovered that early postoperative urine biomarkers including NGAL, IL18, and KIM-1 independently predict LOS, after adjusting for risk and biomarker variables.

KW - Enet

KW - Mnet

KW - negative binomial regression

KW - Poisson regression

KW - Snet

KW - variable selection

UR - http://www.scopus.com/inward/record.url?scp=84995741075&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84995741075&partnerID=8YFLogxK

U2 - 10.1177/0962280214530608

DO - 10.1177/0962280214530608

M3 - Article

C2 - 24742430

AN - SCOPUS:84995741075

VL - 25

SP - 2685

EP - 2703

JO - Statistical Methods in Medical Research

JF - Statistical Methods in Medical Research

SN - 0962-2802

IS - 6

ER -