Adapting machine learning techniques to censored time-to-event health record data: A general-purpose approach using inverse probability of censoring weighting

David M Vock, Julian Wolfson, Sunayan Bandyopadhyay, Gediminas Adomavicius, Paul E. Johnson, Gabriela Vazquez-Benitez, Patrick J. O'Connor

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Models for predicting the probability of experiencing various health outcomes or adverse events over a certain time frame (e.g., having a heart attack in the next 5 years) based on individual patient characteristics are important tools for managing patient care. Electronic health data (EHD) are appealing sources of training data because they provide access to large amounts of rich individual-level data from present-day patient populations. However, because EHD are derived by extracting information from administrative and clinical databases, some fraction of subjects will not be under observation for the entire time frame over which one wants to make predictions; this loss to follow-up is often due to disenrollment from the health system. For subjects without complete follow-up, whether or not they experienced the adverse event is unknown, and in statistical terms the event time is said to be right-censored. Most machine learning approaches to the problem have been relatively ad hoc; for example, common approaches for handling observations in which the event status is unknown include (1) discarding those observations, (2) treating them as non-events, (3) splitting those observations into two observations: one where the event occurs and one where the event does not. In this paper, we present a general-purpose approach to account for right-censored outcomes using inverse probability of censoring weighting (IPCW). We illustrate how IPCW can easily be incorporated into a number of existing machine learning algorithms used to mine big health care data including Bayesian networks, k-nearest neighbors, decision trees, and generalized additive models. We then show that our approach leads to better calibrated predictions than the three ad hoc approaches when applied to predicting the 5-year risk of experiencing a cardiovascular adverse event, using EHD from a large U.S. Midwestern healthcare system.

Original languageEnglish (US)
Pages (from-to)119-131
Number of pages13
JournalJournal of Biomedical Informatics
Volume61
DOIs
StatePublished - Jun 1 2016

Fingerprint

Learning systems
Health
Delivery of Health Care
Decision Trees
Information Storage and Retrieval
Bayesian networks
Decision trees
Health care
Learning algorithms
Patient Care
Myocardial Infarction
Observation
Machine Learning
Databases
Population

Keywords

  • Censored data
  • Electronic health data
  • Inverse probability weighting
  • Machine learning
  • Risk prediction
  • Survival analysis

Cite this

Adapting machine learning techniques to censored time-to-event health record data : A general-purpose approach using inverse probability of censoring weighting. / Vock, David M; Wolfson, Julian; Bandyopadhyay, Sunayan; Adomavicius, Gediminas; Johnson, Paul E.; Vazquez-Benitez, Gabriela; O'Connor, Patrick J.

In: Journal of Biomedical Informatics, Vol. 61, 01.06.2016, p. 119-131.

Research output: Contribution to journalArticle

@article{e58e03ae13204b78b75087bed6f9bf7a,
title = "Adapting machine learning techniques to censored time-to-event health record data: A general-purpose approach using inverse probability of censoring weighting",
abstract = "Models for predicting the probability of experiencing various health outcomes or adverse events over a certain time frame (e.g., having a heart attack in the next 5 years) based on individual patient characteristics are important tools for managing patient care. Electronic health data (EHD) are appealing sources of training data because they provide access to large amounts of rich individual-level data from present-day patient populations. However, because EHD are derived by extracting information from administrative and clinical databases, some fraction of subjects will not be under observation for the entire time frame over which one wants to make predictions; this loss to follow-up is often due to disenrollment from the health system. For subjects without complete follow-up, whether or not they experienced the adverse event is unknown, and in statistical terms the event time is said to be right-censored. Most machine learning approaches to the problem have been relatively ad hoc; for example, common approaches for handling observations in which the event status is unknown include (1) discarding those observations, (2) treating them as non-events, (3) splitting those observations into two observations: one where the event occurs and one where the event does not. In this paper, we present a general-purpose approach to account for right-censored outcomes using inverse probability of censoring weighting (IPCW). We illustrate how IPCW can easily be incorporated into a number of existing machine learning algorithms used to mine big health care data including Bayesian networks, k-nearest neighbors, decision trees, and generalized additive models. We then show that our approach leads to better calibrated predictions than the three ad hoc approaches when applied to predicting the 5-year risk of experiencing a cardiovascular adverse event, using EHD from a large U.S. Midwestern healthcare system.",
keywords = "Censored data, Electronic health data, Inverse probability weighting, Machine learning, Risk prediction, Survival analysis",
author = "Vock, {David M} and Julian Wolfson and Sunayan Bandyopadhyay and Gediminas Adomavicius and Johnson, {Paul E.} and Gabriela Vazquez-Benitez and O'Connor, {Patrick J.}",
year = "2016",
month = "6",
day = "1",
doi = "10.1016/j.jbi.2016.03.009",
language = "English (US)",
volume = "61",
pages = "119--131",
journal = "Journal of Biomedical Informatics",
issn = "1532-0464",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Adapting machine learning techniques to censored time-to-event health record data

T2 - A general-purpose approach using inverse probability of censoring weighting

AU - Vock, David M

AU - Wolfson, Julian

AU - Bandyopadhyay, Sunayan

AU - Adomavicius, Gediminas

AU - Johnson, Paul E.

AU - Vazquez-Benitez, Gabriela

AU - O'Connor, Patrick J.

PY - 2016/6/1

Y1 - 2016/6/1

N2 - Models for predicting the probability of experiencing various health outcomes or adverse events over a certain time frame (e.g., having a heart attack in the next 5 years) based on individual patient characteristics are important tools for managing patient care. Electronic health data (EHD) are appealing sources of training data because they provide access to large amounts of rich individual-level data from present-day patient populations. However, because EHD are derived by extracting information from administrative and clinical databases, some fraction of subjects will not be under observation for the entire time frame over which one wants to make predictions; this loss to follow-up is often due to disenrollment from the health system. For subjects without complete follow-up, whether or not they experienced the adverse event is unknown, and in statistical terms the event time is said to be right-censored. Most machine learning approaches to the problem have been relatively ad hoc; for example, common approaches for handling observations in which the event status is unknown include (1) discarding those observations, (2) treating them as non-events, (3) splitting those observations into two observations: one where the event occurs and one where the event does not. In this paper, we present a general-purpose approach to account for right-censored outcomes using inverse probability of censoring weighting (IPCW). We illustrate how IPCW can easily be incorporated into a number of existing machine learning algorithms used to mine big health care data including Bayesian networks, k-nearest neighbors, decision trees, and generalized additive models. We then show that our approach leads to better calibrated predictions than the three ad hoc approaches when applied to predicting the 5-year risk of experiencing a cardiovascular adverse event, using EHD from a large U.S. Midwestern healthcare system.

AB - Models for predicting the probability of experiencing various health outcomes or adverse events over a certain time frame (e.g., having a heart attack in the next 5 years) based on individual patient characteristics are important tools for managing patient care. Electronic health data (EHD) are appealing sources of training data because they provide access to large amounts of rich individual-level data from present-day patient populations. However, because EHD are derived by extracting information from administrative and clinical databases, some fraction of subjects will not be under observation for the entire time frame over which one wants to make predictions; this loss to follow-up is often due to disenrollment from the health system. For subjects without complete follow-up, whether or not they experienced the adverse event is unknown, and in statistical terms the event time is said to be right-censored. Most machine learning approaches to the problem have been relatively ad hoc; for example, common approaches for handling observations in which the event status is unknown include (1) discarding those observations, (2) treating them as non-events, (3) splitting those observations into two observations: one where the event occurs and one where the event does not. In this paper, we present a general-purpose approach to account for right-censored outcomes using inverse probability of censoring weighting (IPCW). We illustrate how IPCW can easily be incorporated into a number of existing machine learning algorithms used to mine big health care data including Bayesian networks, k-nearest neighbors, decision trees, and generalized additive models. We then show that our approach leads to better calibrated predictions than the three ad hoc approaches when applied to predicting the 5-year risk of experiencing a cardiovascular adverse event, using EHD from a large U.S. Midwestern healthcare system.

KW - Censored data

KW - Electronic health data

KW - Inverse probability weighting

KW - Machine learning

KW - Risk prediction

KW - Survival analysis

UR - http://www.scopus.com/inward/record.url?scp=84962440559&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84962440559&partnerID=8YFLogxK

U2 - 10.1016/j.jbi.2016.03.009

DO - 10.1016/j.jbi.2016.03.009

M3 - Article

C2 - 26992568

AN - SCOPUS:84962440559

VL - 61

SP - 119

EP - 131

JO - Journal of Biomedical Informatics

JF - Journal of Biomedical Informatics

SN - 1532-0464

ER -