Predicting outcomes after liver transplantation

A connectionist approach

Howard Doyle, Igor Dvorchik, Sandi Mitchell, Ignazio R. Marino, Fredrick H. Ebert, John McMichael, John J. Fung

Research output: Contribution to journalArticle

87 Citations (Scopus)

Abstract

Objective: The authors sought to train an artificial neural network to predict early outcomes after orthotopic liver transplantation. Summary Background Data: Reliable prediction of outcomes early after liver transplantation would help improve organ use and could have an impact on patient survival, but remains an elusive goal. Traditional multivariate models have failed to attain the sensitivity and specificity required for practical clinical use. Alternate approaches that can help us model clinical phenomena must be explored. One such approach is the use of artificial neural networks, or connectionist models. These are computation systems that process information in parallel, using large numbers of simple units, and excel in tasks involving pattern recognition. They are capable of adaptive learning and self-organization, and exhibit a high degree of fault tolerance. Methods: Ten feed-forward, back-propagation neural networks were trained to predict graft outcomes, using data from 155 adult liver transplants. The data included information that was available by the second postoperative day. Ten separate training and testing data subsets were prepared, using random sampling, and the ability of the different networks to predict outcomes successfully was evaluated using receiver operating characteristic (ROC) curve analysis. Results: Four of the networks showed perfect discrimination, with an area under the ROC curve (A(z)) of 1.0. Two other networks also had excellent performance, with an A(z) of 0.95. The sensitivity and specificity of the combined networks was 60% and 100%, respectively, when using an output neuron activation of 0.6 as the cutoff point to decide class membership. Lowering the cutoff point to 0.14 increased the sensitivity to 77%, and lowered the specificity to 96%. Conclusions: These results are encouraging, especially when compared to the performance of more traditional multivariate models on the same data set. The robustness of neural networks, when confronted with noisy data generated by nonlinear processes, and their freedom from a priori assumptions regarding the data, make them promising tools with which to develop predictive clinical models.

Original languageEnglish (US)
Pages (from-to)408-415
Number of pages8
JournalAnnals of Surgery
Volume219
Issue number4
StatePublished - Apr 1994
Externally publishedYes

Fingerprint

Neural Networks (Computer)
ROC Curve
Liver Transplantation
Transplants
Sensitivity and Specificity
Aptitude
Information Systems
Learning
Neurons
Survival
Liver
Discrimination (Psychology)
Recognition (Psychology)
Datasets

ASJC Scopus subject areas

  • Surgery

Cite this

Doyle, H., Dvorchik, I., Mitchell, S., Marino, I. R., Ebert, F. H., McMichael, J., & Fung, J. J. (1994). Predicting outcomes after liver transplantation: A connectionist approach. Annals of Surgery, 219(4), 408-415.

Predicting outcomes after liver transplantation : A connectionist approach. / Doyle, Howard; Dvorchik, Igor; Mitchell, Sandi; Marino, Ignazio R.; Ebert, Fredrick H.; McMichael, John; Fung, John J.

In: Annals of Surgery, Vol. 219, No. 4, 04.1994, p. 408-415.

Research output: Contribution to journalArticle

Doyle, H, Dvorchik, I, Mitchell, S, Marino, IR, Ebert, FH, McMichael, J & Fung, JJ 1994, 'Predicting outcomes after liver transplantation: A connectionist approach', Annals of Surgery, vol. 219, no. 4, pp. 408-415.
Doyle H, Dvorchik I, Mitchell S, Marino IR, Ebert FH, McMichael J et al. Predicting outcomes after liver transplantation: A connectionist approach. Annals of Surgery. 1994 Apr;219(4):408-415.
Doyle, Howard ; Dvorchik, Igor ; Mitchell, Sandi ; Marino, Ignazio R. ; Ebert, Fredrick H. ; McMichael, John ; Fung, John J. / Predicting outcomes after liver transplantation : A connectionist approach. In: Annals of Surgery. 1994 ; Vol. 219, No. 4. pp. 408-415.
@article{3ace352f208a46749a9ae008fb3f511b,
title = "Predicting outcomes after liver transplantation: A connectionist approach",
abstract = "Objective: The authors sought to train an artificial neural network to predict early outcomes after orthotopic liver transplantation. Summary Background Data: Reliable prediction of outcomes early after liver transplantation would help improve organ use and could have an impact on patient survival, but remains an elusive goal. Traditional multivariate models have failed to attain the sensitivity and specificity required for practical clinical use. Alternate approaches that can help us model clinical phenomena must be explored. One such approach is the use of artificial neural networks, or connectionist models. These are computation systems that process information in parallel, using large numbers of simple units, and excel in tasks involving pattern recognition. They are capable of adaptive learning and self-organization, and exhibit a high degree of fault tolerance. Methods: Ten feed-forward, back-propagation neural networks were trained to predict graft outcomes, using data from 155 adult liver transplants. The data included information that was available by the second postoperative day. Ten separate training and testing data subsets were prepared, using random sampling, and the ability of the different networks to predict outcomes successfully was evaluated using receiver operating characteristic (ROC) curve analysis. Results: Four of the networks showed perfect discrimination, with an area under the ROC curve (A(z)) of 1.0. Two other networks also had excellent performance, with an A(z) of 0.95. The sensitivity and specificity of the combined networks was 60{\%} and 100{\%}, respectively, when using an output neuron activation of 0.6 as the cutoff point to decide class membership. Lowering the cutoff point to 0.14 increased the sensitivity to 77{\%}, and lowered the specificity to 96{\%}. Conclusions: These results are encouraging, especially when compared to the performance of more traditional multivariate models on the same data set. The robustness of neural networks, when confronted with noisy data generated by nonlinear processes, and their freedom from a priori assumptions regarding the data, make them promising tools with which to develop predictive clinical models.",
author = "Howard Doyle and Igor Dvorchik and Sandi Mitchell and Marino, {Ignazio R.} and Ebert, {Fredrick H.} and John McMichael and Fung, {John J.}",
year = "1994",
month = "4",
language = "English (US)",
volume = "219",
pages = "408--415",
journal = "Annals of Surgery",
issn = "0003-4932",
publisher = "Lippincott Williams and Wilkins",
number = "4",

}

TY - JOUR

T1 - Predicting outcomes after liver transplantation

T2 - A connectionist approach

AU - Doyle, Howard

AU - Dvorchik, Igor

AU - Mitchell, Sandi

AU - Marino, Ignazio R.

AU - Ebert, Fredrick H.

AU - McMichael, John

AU - Fung, John J.

PY - 1994/4

Y1 - 1994/4

N2 - Objective: The authors sought to train an artificial neural network to predict early outcomes after orthotopic liver transplantation. Summary Background Data: Reliable prediction of outcomes early after liver transplantation would help improve organ use and could have an impact on patient survival, but remains an elusive goal. Traditional multivariate models have failed to attain the sensitivity and specificity required for practical clinical use. Alternate approaches that can help us model clinical phenomena must be explored. One such approach is the use of artificial neural networks, or connectionist models. These are computation systems that process information in parallel, using large numbers of simple units, and excel in tasks involving pattern recognition. They are capable of adaptive learning and self-organization, and exhibit a high degree of fault tolerance. Methods: Ten feed-forward, back-propagation neural networks were trained to predict graft outcomes, using data from 155 adult liver transplants. The data included information that was available by the second postoperative day. Ten separate training and testing data subsets were prepared, using random sampling, and the ability of the different networks to predict outcomes successfully was evaluated using receiver operating characteristic (ROC) curve analysis. Results: Four of the networks showed perfect discrimination, with an area under the ROC curve (A(z)) of 1.0. Two other networks also had excellent performance, with an A(z) of 0.95. The sensitivity and specificity of the combined networks was 60% and 100%, respectively, when using an output neuron activation of 0.6 as the cutoff point to decide class membership. Lowering the cutoff point to 0.14 increased the sensitivity to 77%, and lowered the specificity to 96%. Conclusions: These results are encouraging, especially when compared to the performance of more traditional multivariate models on the same data set. The robustness of neural networks, when confronted with noisy data generated by nonlinear processes, and their freedom from a priori assumptions regarding the data, make them promising tools with which to develop predictive clinical models.

AB - Objective: The authors sought to train an artificial neural network to predict early outcomes after orthotopic liver transplantation. Summary Background Data: Reliable prediction of outcomes early after liver transplantation would help improve organ use and could have an impact on patient survival, but remains an elusive goal. Traditional multivariate models have failed to attain the sensitivity and specificity required for practical clinical use. Alternate approaches that can help us model clinical phenomena must be explored. One such approach is the use of artificial neural networks, or connectionist models. These are computation systems that process information in parallel, using large numbers of simple units, and excel in tasks involving pattern recognition. They are capable of adaptive learning and self-organization, and exhibit a high degree of fault tolerance. Methods: Ten feed-forward, back-propagation neural networks were trained to predict graft outcomes, using data from 155 adult liver transplants. The data included information that was available by the second postoperative day. Ten separate training and testing data subsets were prepared, using random sampling, and the ability of the different networks to predict outcomes successfully was evaluated using receiver operating characteristic (ROC) curve analysis. Results: Four of the networks showed perfect discrimination, with an area under the ROC curve (A(z)) of 1.0. Two other networks also had excellent performance, with an A(z) of 0.95. The sensitivity and specificity of the combined networks was 60% and 100%, respectively, when using an output neuron activation of 0.6 as the cutoff point to decide class membership. Lowering the cutoff point to 0.14 increased the sensitivity to 77%, and lowered the specificity to 96%. Conclusions: These results are encouraging, especially when compared to the performance of more traditional multivariate models on the same data set. The robustness of neural networks, when confronted with noisy data generated by nonlinear processes, and their freedom from a priori assumptions regarding the data, make them promising tools with which to develop predictive clinical models.

UR - http://www.scopus.com/inward/record.url?scp=0028261647&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0028261647&partnerID=8YFLogxK

M3 - Article

VL - 219

SP - 408

EP - 415

JO - Annals of Surgery

JF - Annals of Surgery

SN - 0003-4932

IS - 4

ER -