Neural network assessment of perioperative cardiac risk in vascular surgery patients

Pablo Lapuerta, Gilbert J. L'Italien, Sumita Paul, Robert Hendel, Jeffrey A. Leppo, A. Lee Fleisher, Mylan C. Cohen, Kim A. Eagle, Robert P. Giugliano

Research output: Contribution to journalArticle

37 Citations (Scopus)

Abstract

Neural networks were developed to predict perioperative cardiac complications with data from 567 vascular surgery patients. Neural network scores were based on cardiac risk factors and dipyridamole thallium results. These scores were converted into likelihood ratios that predicted cardiac risk. The prognostic accuracy of the neural networks was similar to that of logistic regression models (ROC areas 76.0% vs 75.8%), but their calibration was better. Logistic regression overestimated event rates in a group of high- risk patients (predicted event rate, 64%; observed rate 30%; n = 50, p < 0.001). On a validation set of 514 patients, the neural networks still had ROC similar areas to those of logistic regression (68.3% vs 67.5%), but logistic regression again overestimated event rates for a group of high-risk patients. The calibration difference was reflected in the Hosmer-Lemeshow chi-square statistic (18.8 for the neural networks, 45.0 for logistic regression). The neural networks successfully estimated perioperative cardiac risk with better calibration than comparable logistic regression models.

Original languageEnglish
Pages (from-to)70-75
Number of pages6
JournalMedical Decision Making
Volume18
Issue number1
DOIs
StatePublished - Jan 29 1998
Externally publishedYes

Fingerprint

Blood Vessels
Logistic Models
Calibration
Dipyridamole
Thallium

Keywords

  • Bayes' theorem
  • Cardiac risk
  • Likelihood ratio
  • Logistic regression
  • Neural networks

ASJC Scopus subject areas

  • Public Health, Environmental and Occupational Health
  • Health Informatics
  • Health Information Management
  • Nursing(all)

Cite this

Lapuerta, P., L'Italien, G. J., Paul, S., Hendel, R., Leppo, J. A., Fleisher, A. L., ... Giugliano, R. P. (1998). Neural network assessment of perioperative cardiac risk in vascular surgery patients. Medical Decision Making, 18(1), 70-75. https://doi.org/10.1177/0272989X9801800114

Neural network assessment of perioperative cardiac risk in vascular surgery patients. / Lapuerta, Pablo; L'Italien, Gilbert J.; Paul, Sumita; Hendel, Robert; Leppo, Jeffrey A.; Fleisher, A. Lee; Cohen, Mylan C.; Eagle, Kim A.; Giugliano, Robert P.

In: Medical Decision Making, Vol. 18, No. 1, 29.01.1998, p. 70-75.

Research output: Contribution to journalArticle

Lapuerta, P, L'Italien, GJ, Paul, S, Hendel, R, Leppo, JA, Fleisher, AL, Cohen, MC, Eagle, KA & Giugliano, RP 1998, 'Neural network assessment of perioperative cardiac risk in vascular surgery patients', Medical Decision Making, vol. 18, no. 1, pp. 70-75. https://doi.org/10.1177/0272989X9801800114
Lapuerta P, L'Italien GJ, Paul S, Hendel R, Leppo JA, Fleisher AL et al. Neural network assessment of perioperative cardiac risk in vascular surgery patients. Medical Decision Making. 1998 Jan 29;18(1):70-75. https://doi.org/10.1177/0272989X9801800114
Lapuerta, Pablo ; L'Italien, Gilbert J. ; Paul, Sumita ; Hendel, Robert ; Leppo, Jeffrey A. ; Fleisher, A. Lee ; Cohen, Mylan C. ; Eagle, Kim A. ; Giugliano, Robert P. / Neural network assessment of perioperative cardiac risk in vascular surgery patients. In: Medical Decision Making. 1998 ; Vol. 18, No. 1. pp. 70-75.
@article{5bfb3426b6da428da0d192f233faaeb6,
title = "Neural network assessment of perioperative cardiac risk in vascular surgery patients",
abstract = "Neural networks were developed to predict perioperative cardiac complications with data from 567 vascular surgery patients. Neural network scores were based on cardiac risk factors and dipyridamole thallium results. These scores were converted into likelihood ratios that predicted cardiac risk. The prognostic accuracy of the neural networks was similar to that of logistic regression models (ROC areas 76.0{\%} vs 75.8{\%}), but their calibration was better. Logistic regression overestimated event rates in a group of high- risk patients (predicted event rate, 64{\%}; observed rate 30{\%}; n = 50, p < 0.001). On a validation set of 514 patients, the neural networks still had ROC similar areas to those of logistic regression (68.3{\%} vs 67.5{\%}), but logistic regression again overestimated event rates for a group of high-risk patients. The calibration difference was reflected in the Hosmer-Lemeshow chi-square statistic (18.8 for the neural networks, 45.0 for logistic regression). The neural networks successfully estimated perioperative cardiac risk with better calibration than comparable logistic regression models.",
keywords = "Bayes' theorem, Cardiac risk, Likelihood ratio, Logistic regression, Neural networks",
author = "Pablo Lapuerta and L'Italien, {Gilbert J.} and Sumita Paul and Robert Hendel and Leppo, {Jeffrey A.} and Fleisher, {A. Lee} and Cohen, {Mylan C.} and Eagle, {Kim A.} and Giugliano, {Robert P.}",
year = "1998",
month = "1",
day = "29",
doi = "10.1177/0272989X9801800114",
language = "English",
volume = "18",
pages = "70--75",
journal = "Medical Decision Making",
issn = "0272-989X",
publisher = "SAGE Publications Inc.",
number = "1",

}

TY - JOUR

T1 - Neural network assessment of perioperative cardiac risk in vascular surgery patients

AU - Lapuerta, Pablo

AU - L'Italien, Gilbert J.

AU - Paul, Sumita

AU - Hendel, Robert

AU - Leppo, Jeffrey A.

AU - Fleisher, A. Lee

AU - Cohen, Mylan C.

AU - Eagle, Kim A.

AU - Giugliano, Robert P.

PY - 1998/1/29

Y1 - 1998/1/29

N2 - Neural networks were developed to predict perioperative cardiac complications with data from 567 vascular surgery patients. Neural network scores were based on cardiac risk factors and dipyridamole thallium results. These scores were converted into likelihood ratios that predicted cardiac risk. The prognostic accuracy of the neural networks was similar to that of logistic regression models (ROC areas 76.0% vs 75.8%), but their calibration was better. Logistic regression overestimated event rates in a group of high- risk patients (predicted event rate, 64%; observed rate 30%; n = 50, p < 0.001). On a validation set of 514 patients, the neural networks still had ROC similar areas to those of logistic regression (68.3% vs 67.5%), but logistic regression again overestimated event rates for a group of high-risk patients. The calibration difference was reflected in the Hosmer-Lemeshow chi-square statistic (18.8 for the neural networks, 45.0 for logistic regression). The neural networks successfully estimated perioperative cardiac risk with better calibration than comparable logistic regression models.

AB - Neural networks were developed to predict perioperative cardiac complications with data from 567 vascular surgery patients. Neural network scores were based on cardiac risk factors and dipyridamole thallium results. These scores were converted into likelihood ratios that predicted cardiac risk. The prognostic accuracy of the neural networks was similar to that of logistic regression models (ROC areas 76.0% vs 75.8%), but their calibration was better. Logistic regression overestimated event rates in a group of high- risk patients (predicted event rate, 64%; observed rate 30%; n = 50, p < 0.001). On a validation set of 514 patients, the neural networks still had ROC similar areas to those of logistic regression (68.3% vs 67.5%), but logistic regression again overestimated event rates for a group of high-risk patients. The calibration difference was reflected in the Hosmer-Lemeshow chi-square statistic (18.8 for the neural networks, 45.0 for logistic regression). The neural networks successfully estimated perioperative cardiac risk with better calibration than comparable logistic regression models.

KW - Bayes' theorem

KW - Cardiac risk

KW - Likelihood ratio

KW - Logistic regression

KW - Neural networks

UR - http://www.scopus.com/inward/record.url?scp=0031972710&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031972710&partnerID=8YFLogxK

U2 - 10.1177/0272989X9801800114

DO - 10.1177/0272989X9801800114

M3 - Article

VL - 18

SP - 70

EP - 75

JO - Medical Decision Making

JF - Medical Decision Making

SN - 0272-989X

IS - 1

ER -