A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing

Jonathan Myers, Cesar Roberto De Souza, Audrey Borghi-Silva, Marco Guazzi, Paul Chase, Daniel Bensimhon, Mary Ann Peberdy, Euan Ashley, Erin West, Lawrence P Cahalin, Daniel Forman, Ross Arena

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

Objectives To determine the utility of an artificial neural network (ANN) in predicting cardiovascular (CV) death in patients with heart failure (HF). Background ANNs use weighted inputs in multiple layers of mathematical connections in order to predict outcomes from multiple risk markers. This approach has not been applied in the context of cardiopulmonary exercise testing (CPX) to predict risk in patients with HF. Methods 2635 patients with HF underwent CPX and were followed for a mean of 29 ± 30 months. The sample was divided randomly into ANN training and testing sets to predict CV mortality. Peak VO2, VE/VCO2 slope, heart rate recovery, oxygen uptake efficiency slope, and end-tidal CO2 pressure were included in the model. The predictive accuracy of the ANN was compared to logistic regression (LR) and a Cox proportional hazards (PH) score. A multi-layer feed-forward ANN was used and was tested with a single hidden layer containing a varying number of hidden neurons. Results There were 291 CV deaths during the follow-up. An abnormal VE/VCO2 slope was the strongest predictor of CV mortality using conventional PH analysis (hazard ratio 3.04; 95% CI 2.2-4.2, p < 0.001). After training, the ANN was more accurate in predicting CV mortality compared to LR and PH; ROC areas for the ANN, LR, and PH models were 0.72, 0.70, and 0.69, respectively. Age and BMI-adjusted odds ratios were 4.2, 2.6, and 2.9, for ANN, LR, and PH, respectively. Conclusion An ANN model slightly improves upon conventional methods for estimating CV mortality risk using established CPX responses.

Original languageEnglish
Pages (from-to)265-269
Number of pages5
JournalInternational Journal of Cardiology
Volume171
Issue number2
DOIs
StatePublished - Feb 1 2014

Fingerprint

Heart Failure
Logistic Models
Exercise
Mortality
Neural Networks (Computer)
Proportional Hazards Models
Heart Rate
Odds Ratio
Oxygen
Neurons
Pressure

Keywords

  • Cardiopulmonary exercise testing
  • Heart failure
  • Mortality
  • Oxygen uptake

ASJC Scopus subject areas

  • Cardiology and Cardiovascular Medicine

Cite this

Myers, J., De Souza, C. R., Borghi-Silva, A., Guazzi, M., Chase, P., Bensimhon, D., ... Arena, R. (2014). A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing. International Journal of Cardiology, 171(2), 265-269. https://doi.org/10.1016/j.ijcard.2013.12.031

A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing. / Myers, Jonathan; De Souza, Cesar Roberto; Borghi-Silva, Audrey; Guazzi, Marco; Chase, Paul; Bensimhon, Daniel; Peberdy, Mary Ann; Ashley, Euan; West, Erin; Cahalin, Lawrence P; Forman, Daniel; Arena, Ross.

In: International Journal of Cardiology, Vol. 171, No. 2, 01.02.2014, p. 265-269.

Research output: Contribution to journalArticle

Myers, J, De Souza, CR, Borghi-Silva, A, Guazzi, M, Chase, P, Bensimhon, D, Peberdy, MA, Ashley, E, West, E, Cahalin, LP, Forman, D & Arena, R 2014, 'A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing', International Journal of Cardiology, vol. 171, no. 2, pp. 265-269. https://doi.org/10.1016/j.ijcard.2013.12.031
Myers, Jonathan ; De Souza, Cesar Roberto ; Borghi-Silva, Audrey ; Guazzi, Marco ; Chase, Paul ; Bensimhon, Daniel ; Peberdy, Mary Ann ; Ashley, Euan ; West, Erin ; Cahalin, Lawrence P ; Forman, Daniel ; Arena, Ross. / A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing. In: International Journal of Cardiology. 2014 ; Vol. 171, No. 2. pp. 265-269.
@article{870e85e2cbe845faa364f42e0fe6f25c,
title = "A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing",
abstract = "Objectives To determine the utility of an artificial neural network (ANN) in predicting cardiovascular (CV) death in patients with heart failure (HF). Background ANNs use weighted inputs in multiple layers of mathematical connections in order to predict outcomes from multiple risk markers. This approach has not been applied in the context of cardiopulmonary exercise testing (CPX) to predict risk in patients with HF. Methods 2635 patients with HF underwent CPX and were followed for a mean of 29 ± 30 months. The sample was divided randomly into ANN training and testing sets to predict CV mortality. Peak VO2, VE/VCO2 slope, heart rate recovery, oxygen uptake efficiency slope, and end-tidal CO2 pressure were included in the model. The predictive accuracy of the ANN was compared to logistic regression (LR) and a Cox proportional hazards (PH) score. A multi-layer feed-forward ANN was used and was tested with a single hidden layer containing a varying number of hidden neurons. Results There were 291 CV deaths during the follow-up. An abnormal VE/VCO2 slope was the strongest predictor of CV mortality using conventional PH analysis (hazard ratio 3.04; 95{\%} CI 2.2-4.2, p < 0.001). After training, the ANN was more accurate in predicting CV mortality compared to LR and PH; ROC areas for the ANN, LR, and PH models were 0.72, 0.70, and 0.69, respectively. Age and BMI-adjusted odds ratios were 4.2, 2.6, and 2.9, for ANN, LR, and PH, respectively. Conclusion An ANN model slightly improves upon conventional methods for estimating CV mortality risk using established CPX responses.",
keywords = "Cardiopulmonary exercise testing, Heart failure, Mortality, Oxygen uptake",
author = "Jonathan Myers and {De Souza}, {Cesar Roberto} and Audrey Borghi-Silva and Marco Guazzi and Paul Chase and Daniel Bensimhon and Peberdy, {Mary Ann} and Euan Ashley and Erin West and Cahalin, {Lawrence P} and Daniel Forman and Ross Arena",
year = "2014",
month = "2",
day = "1",
doi = "10.1016/j.ijcard.2013.12.031",
language = "English",
volume = "171",
pages = "265--269",
journal = "International Journal of Cardiology",
issn = "0167-5273",
publisher = "Elsevier Ireland Ltd",
number = "2",

}

TY - JOUR

T1 - A neural network approach to predicting outcomes in heart failure using cardiopulmonary exercise testing

AU - Myers, Jonathan

AU - De Souza, Cesar Roberto

AU - Borghi-Silva, Audrey

AU - Guazzi, Marco

AU - Chase, Paul

AU - Bensimhon, Daniel

AU - Peberdy, Mary Ann

AU - Ashley, Euan

AU - West, Erin

AU - Cahalin, Lawrence P

AU - Forman, Daniel

AU - Arena, Ross

PY - 2014/2/1

Y1 - 2014/2/1

N2 - Objectives To determine the utility of an artificial neural network (ANN) in predicting cardiovascular (CV) death in patients with heart failure (HF). Background ANNs use weighted inputs in multiple layers of mathematical connections in order to predict outcomes from multiple risk markers. This approach has not been applied in the context of cardiopulmonary exercise testing (CPX) to predict risk in patients with HF. Methods 2635 patients with HF underwent CPX and were followed for a mean of 29 ± 30 months. The sample was divided randomly into ANN training and testing sets to predict CV mortality. Peak VO2, VE/VCO2 slope, heart rate recovery, oxygen uptake efficiency slope, and end-tidal CO2 pressure were included in the model. The predictive accuracy of the ANN was compared to logistic regression (LR) and a Cox proportional hazards (PH) score. A multi-layer feed-forward ANN was used and was tested with a single hidden layer containing a varying number of hidden neurons. Results There were 291 CV deaths during the follow-up. An abnormal VE/VCO2 slope was the strongest predictor of CV mortality using conventional PH analysis (hazard ratio 3.04; 95% CI 2.2-4.2, p < 0.001). After training, the ANN was more accurate in predicting CV mortality compared to LR and PH; ROC areas for the ANN, LR, and PH models were 0.72, 0.70, and 0.69, respectively. Age and BMI-adjusted odds ratios were 4.2, 2.6, and 2.9, for ANN, LR, and PH, respectively. Conclusion An ANN model slightly improves upon conventional methods for estimating CV mortality risk using established CPX responses.

AB - Objectives To determine the utility of an artificial neural network (ANN) in predicting cardiovascular (CV) death in patients with heart failure (HF). Background ANNs use weighted inputs in multiple layers of mathematical connections in order to predict outcomes from multiple risk markers. This approach has not been applied in the context of cardiopulmonary exercise testing (CPX) to predict risk in patients with HF. Methods 2635 patients with HF underwent CPX and were followed for a mean of 29 ± 30 months. The sample was divided randomly into ANN training and testing sets to predict CV mortality. Peak VO2, VE/VCO2 slope, heart rate recovery, oxygen uptake efficiency slope, and end-tidal CO2 pressure were included in the model. The predictive accuracy of the ANN was compared to logistic regression (LR) and a Cox proportional hazards (PH) score. A multi-layer feed-forward ANN was used and was tested with a single hidden layer containing a varying number of hidden neurons. Results There were 291 CV deaths during the follow-up. An abnormal VE/VCO2 slope was the strongest predictor of CV mortality using conventional PH analysis (hazard ratio 3.04; 95% CI 2.2-4.2, p < 0.001). After training, the ANN was more accurate in predicting CV mortality compared to LR and PH; ROC areas for the ANN, LR, and PH models were 0.72, 0.70, and 0.69, respectively. Age and BMI-adjusted odds ratios were 4.2, 2.6, and 2.9, for ANN, LR, and PH, respectively. Conclusion An ANN model slightly improves upon conventional methods for estimating CV mortality risk using established CPX responses.

KW - Cardiopulmonary exercise testing

KW - Heart failure

KW - Mortality

KW - Oxygen uptake

UR - http://www.scopus.com/inward/record.url?scp=84892869562&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84892869562&partnerID=8YFLogxK

U2 - 10.1016/j.ijcard.2013.12.031

DO - 10.1016/j.ijcard.2013.12.031

M3 - Article

VL - 171

SP - 265

EP - 269

JO - International Journal of Cardiology

JF - International Journal of Cardiology

SN - 0167-5273

IS - 2

ER -