Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines

JD Diaz, OJ Escalona, NC Castro, JMCC Anderson, B Glover, G Manoharan

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

1 Citation (Scopus)

Abstract

In this work, we use support vector machines (SVM) to predict if a defibrillation shock is likely to be successful or not in the cardioversion of persistent AF patients. TheECG signals of 47 patients elected for electrical cardioversion treatment were collected at the Royal Victoria Hospital in Belfast city, NI-UK. Signal processing was performed on ECG segments prior each shock. Three electrocardiographic indexes were extracted and used as input: the dominant atrial fibrillatory frequency, the mean and the standard deviation of the R-R interval time series of the ECG segments. We trained SVM using about 40% of the data. SVM could predict the outcome of 89% of low-energy shocks below or = 100 [J], with a sensitivity (SE) of 87.50% and specificity (SP) of 98.8%. As a remarkable result, theoutcome of higher energy shocks (above or = 150 [J]) could be predicted with 100% exactitude.
LanguageEnglish
Title of host publicationUnknown Host Publication
Place of Publicationwww
Pages741-744
Number of pages4
Volume37
Publication statusPublished - 15 Nov 2010
EventComputing in Cardiology - Belfast-UK
Duration: 15 Nov 2010 → …
http://www.cinc.org

Conference

ConferenceComputing in Cardiology
Period15/11/10 → …
Internet address

Fingerprint

Electric Countershock
Atrial Fibrillation
Shock
Electrocardiography
Victoria
Sensitivity and Specificity
Support Vector Machine
Therapeutics

Keywords

  • AF Cardioversion
  • Atrial Fibrillation
  • AF Defibrillation
  • Support Vector Machine
  • Arrhythmia Treatment
  • ECG Time Series
  • QRS Cancellation

Cite this

Diaz, JD., Escalona, OJ., Castro, NC., Anderson, JMCC., Glover, B., & Manoharan, G. (2010). Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines. In Unknown Host Publication (Vol. 37, pp. 741-744). www.
Diaz, JD ; Escalona, OJ ; Castro, NC ; Anderson, JMCC ; Glover, B ; Manoharan, G. / Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines. Unknown Host Publication. Vol. 37 www, 2010. pp. 741-744
@inproceedings{1af58b0ef01940eb9bf27269e26e10b2,
title = "Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines",
abstract = "In this work, we use support vector machines (SVM) to predict if a defibrillation shock is likely to be successful or not in the cardioversion of persistent AF patients. TheECG signals of 47 patients elected for electrical cardioversion treatment were collected at the Royal Victoria Hospital in Belfast city, NI-UK. Signal processing was performed on ECG segments prior each shock. Three electrocardiographic indexes were extracted and used as input: the dominant atrial fibrillatory frequency, the mean and the standard deviation of the R-R interval time series of the ECG segments. We trained SVM using about 40{\%} of the data. SVM could predict the outcome of 89{\%} of low-energy shocks below or = 100 [J], with a sensitivity (SE) of 87.50{\%} and specificity (SP) of 98.8{\%}. As a remarkable result, theoutcome of higher energy shocks (above or = 150 [J]) could be predicted with 100{\%} exactitude.",
keywords = "AF Cardioversion, Atrial Fibrillation, AF Defibrillation, Support Vector Machine, Arrhythmia Treatment, ECG Time Series, QRS Cancellation",
author = "JD Diaz and OJ Escalona and NC Castro and JMCC Anderson and B Glover and G Manoharan",
year = "2010",
month = "11",
day = "15",
language = "English",
volume = "37",
pages = "741--744",
booktitle = "Unknown Host Publication",

}

Diaz, JD, Escalona, OJ, Castro, NC, Anderson, JMCC, Glover, B & Manoharan, G 2010, Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines. in Unknown Host Publication. vol. 37, www, pp. 741-744, Computing in Cardiology, 15/11/10.

Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines. / Diaz, JD; Escalona, OJ; Castro, NC; Anderson, JMCC; Glover, B; Manoharan, G.

Unknown Host Publication. Vol. 37 www, 2010. p. 741-744.

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

TY - GEN

T1 - Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines

AU - Diaz, JD

AU - Escalona, OJ

AU - Castro, NC

AU - Anderson, JMCC

AU - Glover, B

AU - Manoharan, G

PY - 2010/11/15

Y1 - 2010/11/15

N2 - In this work, we use support vector machines (SVM) to predict if a defibrillation shock is likely to be successful or not in the cardioversion of persistent AF patients. TheECG signals of 47 patients elected for electrical cardioversion treatment were collected at the Royal Victoria Hospital in Belfast city, NI-UK. Signal processing was performed on ECG segments prior each shock. Three electrocardiographic indexes were extracted and used as input: the dominant atrial fibrillatory frequency, the mean and the standard deviation of the R-R interval time series of the ECG segments. We trained SVM using about 40% of the data. SVM could predict the outcome of 89% of low-energy shocks below or = 100 [J], with a sensitivity (SE) of 87.50% and specificity (SP) of 98.8%. As a remarkable result, theoutcome of higher energy shocks (above or = 150 [J]) could be predicted with 100% exactitude.

AB - In this work, we use support vector machines (SVM) to predict if a defibrillation shock is likely to be successful or not in the cardioversion of persistent AF patients. TheECG signals of 47 patients elected for electrical cardioversion treatment were collected at the Royal Victoria Hospital in Belfast city, NI-UK. Signal processing was performed on ECG segments prior each shock. Three electrocardiographic indexes were extracted and used as input: the dominant atrial fibrillatory frequency, the mean and the standard deviation of the R-R interval time series of the ECG segments. We trained SVM using about 40% of the data. SVM could predict the outcome of 89% of low-energy shocks below or = 100 [J], with a sensitivity (SE) of 87.50% and specificity (SP) of 98.8%. As a remarkable result, theoutcome of higher energy shocks (above or = 150 [J]) could be predicted with 100% exactitude.

KW - AF Cardioversion

KW - Atrial Fibrillation

KW - AF Defibrillation

KW - Support Vector Machine

KW - Arrhythmia Treatment

KW - ECG Time Series

KW - QRS Cancellation

M3 - Conference contribution

VL - 37

SP - 741

EP - 744

BT - Unknown Host Publication

CY - www

ER -

Diaz JD, Escalona OJ, Castro NC, Anderson JMCC, Glover B, Manoharan G. Predicting Transthoracic Defibrillation Shocks Outcome in the Cardioversion of Atrial Fibrillation Employing Support Vector Machines. In Unknown Host Publication. Vol. 37. www. 2010. p. 741-744