Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

Abstract

Cardiac patients undergoing surgery face increased risk of postoperative complications, due to a combination of factors, including higher risk surgery, their age at time of surgery and the presence of co-morbid conditions. They will therefore require high levels of care and clinical resources throughout their perioperative journey (i.e. before, during and after surgery). Although surgical mortality rates in the UK have remained low, postoperative complications on the other hand are common and can have a significant impact on patients’ quality of life, increase hospital length of stay and healthcare costs. In this study we used and compared several machine learning methods – random forest, AdaBoost, gradient boosting model and stacking – to predict severe postoperative complications after cardiac surgery based on preoperative variables obtained from a surgical database of a large acute care hospital in Scotland. Our results show that AdaBoost has the best overall performance (AUC = 0.731), and also outperforms EuroSCORE and EuroSCORE II in other studies predicting postoperative complications. Random forest (Sensitivity = 0.852, negative predictive value = 0.923), however, and gradient boosting model (Sensitivity = 0.875 and negative predictive value = 0.920) have the best performance at predicting severe postoperative complications based on sensitivity and negative predictive value.

Original languageEnglish
Title of host publicationArtificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings
EditorsSzymon Wilk, Annette ten Teije, David Riaño
Place of PublicationCham
PublisherSpringer
Pages376–385
Number of pages10
ISBN (Electronic)9783030216429
ISBN (Print)9783030216412
DOIs
Publication statusPublished - 30 May 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11526 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Fingerprint

Surgery
Adaptive boosting
Learning systems
Costs

Keywords

  • postoperative complications
  • machine learning
  • cardiac surgery

Cite this

Lapp, L., Bouamrane, M-M., Kavanagh, K., Roper, M., Young, D., & Schraag, S. (2019). Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery. In S. Wilk, A. ten Teije, & D. Riaño (Eds.), Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings (pp. 376–385). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11526 LNAI). Cham: Springer. https://doi.org/10.1007/978-3-030-21642-9_48
Lapp, Linda ; Bouamrane, Matt-Mouley ; Kavanagh, Kimberley ; Roper, Marc ; Young, David ; Schraag, Stefan. / Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery. Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings. editor / Szymon Wilk ; Annette ten Teije ; David Riaño. Cham : Springer, 2019. pp. 376–385 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{b19f214b0007437c8fe735d2dbaa01f7,
title = "Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery",
abstract = "Cardiac patients undergoing surgery face increased risk of postoperative complications, due to a combination of factors, including higher risk surgery, their age at time of surgery and the presence of co-morbid conditions. They will therefore require high levels of care and clinical resources throughout their perioperative journey (i.e. before, during and after surgery). Although surgical mortality rates in the UK have remained low, postoperative complications on the other hand are common and can have a significant impact on patients’ quality of life, increase hospital length of stay and healthcare costs. In this study we used and compared several machine learning methods – random forest, AdaBoost, gradient boosting model and stacking – to predict severe postoperative complications after cardiac surgery based on preoperative variables obtained from a surgical database of a large acute care hospital in Scotland. Our results show that AdaBoost has the best overall performance (AUC = 0.731), and also outperforms EuroSCORE and EuroSCORE II in other studies predicting postoperative complications. Random forest (Sensitivity = 0.852, negative predictive value = 0.923), however, and gradient boosting model (Sensitivity = 0.875 and negative predictive value = 0.920) have the best performance at predicting severe postoperative complications based on sensitivity and negative predictive value.",
keywords = "postoperative complications, machine learning, cardiac surgery",
author = "Linda Lapp and Matt-Mouley Bouamrane and Kimberley Kavanagh and Marc Roper and David Young and Stefan Schraag",
note = "This is a post-peer-review, pre-copyedit version of an article published in Artificial Intelligence in Medicine. The final authenticated version is available online at: http://dx.doi.org/10.1007/978-3-030-21642-9_48.",
year = "2019",
month = "5",
day = "30",
doi = "10.1007/978-3-030-21642-9_48",
language = "English",
isbn = "9783030216412",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer",
pages = "376–385",
editor = "Szymon Wilk and {ten Teije}, Annette and David Ria{\~n}o",
booktitle = "Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings",

}

Lapp, L, Bouamrane, M-M, Kavanagh, K, Roper, M, Young, D & Schraag, S 2019, Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery. in S Wilk, A ten Teije & D Riaño (eds), Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11526 LNAI, Springer, Cham, pp. 376–385. https://doi.org/10.1007/978-3-030-21642-9_48

Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery. / Lapp, Linda; Bouamrane, Matt-Mouley; Kavanagh, Kimberley; Roper, Marc; Young, David; Schraag, Stefan.

Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings. ed. / Szymon Wilk; Annette ten Teije; David Riaño. Cham : Springer, 2019. p. 376–385 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11526 LNAI).

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

TY - GEN

T1 - Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery

AU - Lapp, Linda

AU - Bouamrane, Matt-Mouley

AU - Kavanagh, Kimberley

AU - Roper, Marc

AU - Young, David

AU - Schraag, Stefan

N1 - This is a post-peer-review, pre-copyedit version of an article published in Artificial Intelligence in Medicine. The final authenticated version is available online at: http://dx.doi.org/10.1007/978-3-030-21642-9_48.

PY - 2019/5/30

Y1 - 2019/5/30

N2 - Cardiac patients undergoing surgery face increased risk of postoperative complications, due to a combination of factors, including higher risk surgery, their age at time of surgery and the presence of co-morbid conditions. They will therefore require high levels of care and clinical resources throughout their perioperative journey (i.e. before, during and after surgery). Although surgical mortality rates in the UK have remained low, postoperative complications on the other hand are common and can have a significant impact on patients’ quality of life, increase hospital length of stay and healthcare costs. In this study we used and compared several machine learning methods – random forest, AdaBoost, gradient boosting model and stacking – to predict severe postoperative complications after cardiac surgery based on preoperative variables obtained from a surgical database of a large acute care hospital in Scotland. Our results show that AdaBoost has the best overall performance (AUC = 0.731), and also outperforms EuroSCORE and EuroSCORE II in other studies predicting postoperative complications. Random forest (Sensitivity = 0.852, negative predictive value = 0.923), however, and gradient boosting model (Sensitivity = 0.875 and negative predictive value = 0.920) have the best performance at predicting severe postoperative complications based on sensitivity and negative predictive value.

AB - Cardiac patients undergoing surgery face increased risk of postoperative complications, due to a combination of factors, including higher risk surgery, their age at time of surgery and the presence of co-morbid conditions. They will therefore require high levels of care and clinical resources throughout their perioperative journey (i.e. before, during and after surgery). Although surgical mortality rates in the UK have remained low, postoperative complications on the other hand are common and can have a significant impact on patients’ quality of life, increase hospital length of stay and healthcare costs. In this study we used and compared several machine learning methods – random forest, AdaBoost, gradient boosting model and stacking – to predict severe postoperative complications after cardiac surgery based on preoperative variables obtained from a surgical database of a large acute care hospital in Scotland. Our results show that AdaBoost has the best overall performance (AUC = 0.731), and also outperforms EuroSCORE and EuroSCORE II in other studies predicting postoperative complications. Random forest (Sensitivity = 0.852, negative predictive value = 0.923), however, and gradient boosting model (Sensitivity = 0.875 and negative predictive value = 0.920) have the best performance at predicting severe postoperative complications based on sensitivity and negative predictive value.

KW - postoperative complications

KW - machine learning

KW - cardiac surgery

U2 - 10.1007/978-3-030-21642-9_48

DO - 10.1007/978-3-030-21642-9_48

M3 - Conference contribution book

SN - 9783030216412

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 376

EP - 385

BT - Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings

A2 - Wilk, Szymon

A2 - ten Teije, Annette

A2 - Riaño, David

PB - Springer

CY - Cham

ER -

Lapp L, Bouamrane M-M, Kavanagh K, Roper M, Young D, Schraag S. Evaluation of random forest and ensemble methods at predicting complications following cardiac surgery. In Wilk S, ten Teije A, Riaño D, editors, Artificial Intelligence in Medicine - 17th Conference on Artificial Intelligence in Medicine, AIME 2019, Proceedings. Cham: Springer. 2019. p. 376–385. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-21642-9_48