A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest

Nilesh Pareek; Christopher Frohmaier; Mathew Smith; Peter Kordis; Antonio Cannata; Jo Nevett; Rachael Fothergill; Robert C Nichol; Mark Sullivan; Nicholas Sunderland; Thomas W Johnson; Marko Noc; Jonathan Byrne; Philip MacCarthy; Ajay M Shah

doi:10.1002/ccd.30677

A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest

Nilesh Pareek^*, Christopher Frohmaier, Mathew Smith, Peter Kordis, Antonio Cannata, Jo Nevett, Rachael Fothergill, Robert C Nichol, Mark Sullivan, Nicholas Sunderland, Thomas W Johnson, Marko Noc, Jonathan Byrne, Philip MacCarthy, Ajay M Shah

^*Corresponding author for this work

Bristol Medical School (THS)

Research output: Contribution to journal › Article (Academic Journal) › peer-review

3 Citations (Scopus)

Abstract

BACKGROUND: We aimed to develop a machine learning algorithm to predict the presence of a culprit lesion in patients with out-of-hospital cardiac arrest (OHCA).

METHODS: We used the King's Out-of-Hospital Cardiac Arrest Registry, a retrospective cohort of 398 patients admitted to King's College Hospital between May 2012 and December 2017. The primary outcome was the presence of a culprit coronary artery lesion, for which a gradient boosting model was optimized to predict. The algorithm was then validated in two independent European cohorts comprising 568 patients.

RESULTS: A culprit lesion was observed in 209/309 (67.4%) patients receiving early coronary angiography in the development, and 199/293 (67.9%) in the Ljubljana and 102/132 (61.1%) in the Bristol validation cohorts, respectively. The algorithm, which is presented as a web application, incorporates nine variables including age, a localizing feature on electrocardiogram (ECG) (≥2 mm of ST change in contiguous leads), regional wall motion abnormality, history of vascular disease and initial shockable rhythm. This model had an area under the curve (AUC) of 0.89 in the development and 0.83/0.81 in the validation cohorts with good calibration and outperforms the current gold standard-ECG alone (AUC: 0.69/0.67/0/67).

CONCLUSIONS: A novel simple machine learning-derived algorithm can be applied to patients with OHCA, to predict a culprit coronary artery disease lesion with high accuracy.

Original language	English
Pages (from-to)	80-90
Number of pages	11
Journal	Catheterization and Cardiovascular Interventions
Volume	102
Issue number	1
Early online date	16 May 2023
DOIs	https://doi.org/10.1002/ccd.30677
Publication status	Published - 1 Jul 2023

Bibliographical note

Publisher Copyright:
© 2023 The Authors. Catheterization and Cardiovascular Interventions published by Wiley Periodicals LLC.

Keywords

Humans
Out-of-Hospital Cardiac Arrest/diagnosis
Retrospective Studies
Treatment Outcome
Coronary Artery Disease/diagnostic imaging
Coronary Angiography
Algorithms
Cardiopulmonary Resuscitation

Access to Document

10.1002/ccd.30677Licence: CC BY-NC-ND

Handle.net

Persistent link

Cite this

Pareek, N., Frohmaier, C., Smith, M., Kordis, P., Cannata, A., Nevett, J., Fothergill, R., Nichol, R. C., Sullivan, M., Sunderland, N., Johnson, T. W., Noc, M., Byrne, J., MacCarthy, P., & Shah, A. M. (2023). A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest. Catheterization and Cardiovascular Interventions, 102(1), 80-90. https://doi.org/10.1002/ccd.30677

@article{1a287fac216943d5812b6d1a9f22fdc8,

title = "A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest",

abstract = "BACKGROUND: We aimed to develop a machine learning algorithm to predict the presence of a culprit lesion in patients with out-of-hospital cardiac arrest (OHCA).METHODS: We used the King's Out-of-Hospital Cardiac Arrest Registry, a retrospective cohort of 398 patients admitted to King's College Hospital between May 2012 and December 2017. The primary outcome was the presence of a culprit coronary artery lesion, for which a gradient boosting model was optimized to predict. The algorithm was then validated in two independent European cohorts comprising 568 patients.RESULTS: A culprit lesion was observed in 209/309 (67.4%) patients receiving early coronary angiography in the development, and 199/293 (67.9%) in the Ljubljana and 102/132 (61.1%) in the Bristol validation cohorts, respectively. The algorithm, which is presented as a web application, incorporates nine variables including age, a localizing feature on electrocardiogram (ECG) (≥2 mm of ST change in contiguous leads), regional wall motion abnormality, history of vascular disease and initial shockable rhythm. This model had an area under the curve (AUC) of 0.89 in the development and 0.83/0.81 in the validation cohorts with good calibration and outperforms the current gold standard-ECG alone (AUC: 0.69/0.67/0/67).CONCLUSIONS: A novel simple machine learning-derived algorithm can be applied to patients with OHCA, to predict a culprit coronary artery disease lesion with high accuracy.",

keywords = "Humans, Out-of-Hospital Cardiac Arrest/diagnosis, Retrospective Studies, Treatment Outcome, Coronary Artery Disease/diagnostic imaging, Coronary Angiography, Algorithms, Cardiopulmonary Resuscitation",

author = "Nilesh Pareek and Christopher Frohmaier and Mathew Smith and Peter Kordis and Antonio Cannata and Jo Nevett and Rachael Fothergill and Nichol, {Robert C} and Mark Sullivan and Nicholas Sunderland and Johnson, {Thomas W} and Marko Noc and Jonathan Byrne and Philip MacCarthy and Shah, {Ajay M}",

note = "Publisher Copyright: {\textcopyright} 2023 The Authors. Catheterization and Cardiovascular Interventions published by Wiley Periodicals LLC.",

year = "2023",

month = jul,

day = "1",

doi = "10.1002/ccd.30677",

language = "English",

volume = "102",

pages = "80--90",

journal = "Catheterization and Cardiovascular Interventions",

issn = "1522-1946",

publisher = "Wiley",

number = "1",

}

Pareek, N, Frohmaier, C, Smith, M, Kordis, P, Cannata, A, Nevett, J, Fothergill, R, Nichol, RC, Sullivan, M, Sunderland, N , Johnson, TW, Noc, M, Byrne, J, MacCarthy, P & Shah, AM 2023, 'A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest', Catheterization and Cardiovascular Interventions, vol. 102, no. 1, pp. 80-90. https://doi.org/10.1002/ccd.30677

TY - JOUR

T1 - A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest

AU - Pareek, Nilesh

AU - Frohmaier, Christopher

AU - Smith, Mathew

AU - Kordis, Peter

AU - Cannata, Antonio

AU - Nevett, Jo

AU - Fothergill, Rachael

AU - Nichol, Robert C

AU - Sullivan, Mark

AU - Sunderland, Nicholas

AU - Johnson, Thomas W

AU - Noc, Marko

AU - Byrne, Jonathan

AU - MacCarthy, Philip

AU - Shah, Ajay M

PY - 2023/7/1

Y1 - 2023/7/1

N2 - BACKGROUND: We aimed to develop a machine learning algorithm to predict the presence of a culprit lesion in patients with out-of-hospital cardiac arrest (OHCA).METHODS: We used the King's Out-of-Hospital Cardiac Arrest Registry, a retrospective cohort of 398 patients admitted to King's College Hospital between May 2012 and December 2017. The primary outcome was the presence of a culprit coronary artery lesion, for which a gradient boosting model was optimized to predict. The algorithm was then validated in two independent European cohorts comprising 568 patients.RESULTS: A culprit lesion was observed in 209/309 (67.4%) patients receiving early coronary angiography in the development, and 199/293 (67.9%) in the Ljubljana and 102/132 (61.1%) in the Bristol validation cohorts, respectively. The algorithm, which is presented as a web application, incorporates nine variables including age, a localizing feature on electrocardiogram (ECG) (≥2 mm of ST change in contiguous leads), regional wall motion abnormality, history of vascular disease and initial shockable rhythm. This model had an area under the curve (AUC) of 0.89 in the development and 0.83/0.81 in the validation cohorts with good calibration and outperforms the current gold standard-ECG alone (AUC: 0.69/0.67/0/67).CONCLUSIONS: A novel simple machine learning-derived algorithm can be applied to patients with OHCA, to predict a culprit coronary artery disease lesion with high accuracy.

AB - BACKGROUND: We aimed to develop a machine learning algorithm to predict the presence of a culprit lesion in patients with out-of-hospital cardiac arrest (OHCA).METHODS: We used the King's Out-of-Hospital Cardiac Arrest Registry, a retrospective cohort of 398 patients admitted to King's College Hospital between May 2012 and December 2017. The primary outcome was the presence of a culprit coronary artery lesion, for which a gradient boosting model was optimized to predict. The algorithm was then validated in two independent European cohorts comprising 568 patients.RESULTS: A culprit lesion was observed in 209/309 (67.4%) patients receiving early coronary angiography in the development, and 199/293 (67.9%) in the Ljubljana and 102/132 (61.1%) in the Bristol validation cohorts, respectively. The algorithm, which is presented as a web application, incorporates nine variables including age, a localizing feature on electrocardiogram (ECG) (≥2 mm of ST change in contiguous leads), regional wall motion abnormality, history of vascular disease and initial shockable rhythm. This model had an area under the curve (AUC) of 0.89 in the development and 0.83/0.81 in the validation cohorts with good calibration and outperforms the current gold standard-ECG alone (AUC: 0.69/0.67/0/67).CONCLUSIONS: A novel simple machine learning-derived algorithm can be applied to patients with OHCA, to predict a culprit coronary artery disease lesion with high accuracy.

KW - Humans

KW - Out-of-Hospital Cardiac Arrest/diagnosis

KW - Retrospective Studies

KW - Treatment Outcome

KW - Coronary Artery Disease/diagnostic imaging

KW - Coronary Angiography

KW - Algorithms

KW - Cardiopulmonary Resuscitation

U2 - 10.1002/ccd.30677

DO - 10.1002/ccd.30677

M3 - Article (Academic Journal)

C2 - 37191312

SN - 1522-1946

VL - 102

SP - 80

EP - 90

JO - Catheterization and Cardiovascular Interventions

JF - Catheterization and Cardiovascular Interventions

IS - 1

ER -

A machine learning algorithm to predict a culprit lesion after out of hospital cardiac arrest

Abstract

Bibliographical note

Keywords

Access to Document

Handle.net

Fingerprint

Cite this