Moving beyond regression techniques in cardiovascular risk prediction

Applying machine learning to address analytic challenges

Benjamin A. Goldstein, Ann Marie Navar, Rickey E. Carter

Research output: Contribution to journalReview article

43 Citations (Scopus)

Abstract

Risk prediction plays an important role in clinical cardiology research. Traditionally, most risk models have been based on regression models. While useful and robust, these statistical methods are limited to using a small number of predictors which operate in the sameway on everyone, and uniformly throughout their range. The purpose of this review is to illustrate the use of machine-learning methods for development of risk prediction models. Typically presented as black box approaches, most machine-learning methods are aimed at solving particular challenges that arise in data analysis that are not well addressed by typical regression approaches. To illustrate these challenges, as well as how different methods can address them, we consider trying to predicting mortality after diagnosis of acute myocardial infarction. We use data derived from our institution's electronic health record and abstract data on 13 regularly measured laboratory markers. We walk through different challenges that arise in modelling these data and then introduce different machine-learning approaches. Finally, we discuss general issues in the application of machine-learning methods including tuning parameters, loss functions, variable importance, and missing data. Overall, this review serves as an introduction for those working on risk modelling to approach the diffuse field of machine learning.

Original languageEnglish (US)
Pages (from-to)1805-1814
Number of pages10
JournalEuropean Heart Journal
Volume38
Issue number23
DOIs
StatePublished - Jun 14 2017

Fingerprint

Electronic Health Records
Cardiology
Biomarkers
Myocardial Infarction
Machine Learning
Mortality
Research

Keywords

  • Electronic health records
  • Personalized medicine
  • Precision medicine
  • Risk prediction

ASJC Scopus subject areas

  • Cardiology and Cardiovascular Medicine

Cite this

Moving beyond regression techniques in cardiovascular risk prediction : Applying machine learning to address analytic challenges. / Goldstein, Benjamin A.; Navar, Ann Marie; Carter, Rickey E.

In: European Heart Journal, Vol. 38, No. 23, 14.06.2017, p. 1805-1814.

Research output: Contribution to journalReview article

@article{8bfb643685594d03ba25dd900fbac910,
title = "Moving beyond regression techniques in cardiovascular risk prediction: Applying machine learning to address analytic challenges",
abstract = "Risk prediction plays an important role in clinical cardiology research. Traditionally, most risk models have been based on regression models. While useful and robust, these statistical methods are limited to using a small number of predictors which operate in the sameway on everyone, and uniformly throughout their range. The purpose of this review is to illustrate the use of machine-learning methods for development of risk prediction models. Typically presented as black box approaches, most machine-learning methods are aimed at solving particular challenges that arise in data analysis that are not well addressed by typical regression approaches. To illustrate these challenges, as well as how different methods can address them, we consider trying to predicting mortality after diagnosis of acute myocardial infarction. We use data derived from our institution's electronic health record and abstract data on 13 regularly measured laboratory markers. We walk through different challenges that arise in modelling these data and then introduce different machine-learning approaches. Finally, we discuss general issues in the application of machine-learning methods including tuning parameters, loss functions, variable importance, and missing data. Overall, this review serves as an introduction for those working on risk modelling to approach the diffuse field of machine learning.",
keywords = "Electronic health records, Personalized medicine, Precision medicine, Risk prediction",
author = "Goldstein, {Benjamin A.} and Navar, {Ann Marie} and Carter, {Rickey E.}",
year = "2017",
month = "6",
day = "14",
doi = "10.1093/eurheartj/ehw302",
language = "English (US)",
volume = "38",
pages = "1805--1814",
journal = "European Heart Journal",
issn = "0195-668X",
publisher = "Oxford University Press",
number = "23",

}

TY - JOUR

T1 - Moving beyond regression techniques in cardiovascular risk prediction

T2 - Applying machine learning to address analytic challenges

AU - Goldstein, Benjamin A.

AU - Navar, Ann Marie

AU - Carter, Rickey E.

PY - 2017/6/14

Y1 - 2017/6/14

N2 - Risk prediction plays an important role in clinical cardiology research. Traditionally, most risk models have been based on regression models. While useful and robust, these statistical methods are limited to using a small number of predictors which operate in the sameway on everyone, and uniformly throughout their range. The purpose of this review is to illustrate the use of machine-learning methods for development of risk prediction models. Typically presented as black box approaches, most machine-learning methods are aimed at solving particular challenges that arise in data analysis that are not well addressed by typical regression approaches. To illustrate these challenges, as well as how different methods can address them, we consider trying to predicting mortality after diagnosis of acute myocardial infarction. We use data derived from our institution's electronic health record and abstract data on 13 regularly measured laboratory markers. We walk through different challenges that arise in modelling these data and then introduce different machine-learning approaches. Finally, we discuss general issues in the application of machine-learning methods including tuning parameters, loss functions, variable importance, and missing data. Overall, this review serves as an introduction for those working on risk modelling to approach the diffuse field of machine learning.

AB - Risk prediction plays an important role in clinical cardiology research. Traditionally, most risk models have been based on regression models. While useful and robust, these statistical methods are limited to using a small number of predictors which operate in the sameway on everyone, and uniformly throughout their range. The purpose of this review is to illustrate the use of machine-learning methods for development of risk prediction models. Typically presented as black box approaches, most machine-learning methods are aimed at solving particular challenges that arise in data analysis that are not well addressed by typical regression approaches. To illustrate these challenges, as well as how different methods can address them, we consider trying to predicting mortality after diagnosis of acute myocardial infarction. We use data derived from our institution's electronic health record and abstract data on 13 regularly measured laboratory markers. We walk through different challenges that arise in modelling these data and then introduce different machine-learning approaches. Finally, we discuss general issues in the application of machine-learning methods including tuning parameters, loss functions, variable importance, and missing data. Overall, this review serves as an introduction for those working on risk modelling to approach the diffuse field of machine learning.

KW - Electronic health records

KW - Personalized medicine

KW - Precision medicine

KW - Risk prediction

UR - http://www.scopus.com/inward/record.url?scp=85021953748&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85021953748&partnerID=8YFLogxK

U2 - 10.1093/eurheartj/ehw302

DO - 10.1093/eurheartj/ehw302

M3 - Review article

VL - 38

SP - 1805

EP - 1814

JO - European Heart Journal

JF - European Heart Journal

SN - 0195-668X

IS - 23

ER -