Building clinical classifiers using incomplete observations - A neural network ensemble for hepatoma detection in patients with cirrhosis

H. R. Doyle, B. Parmanto, P. W. Munro, I. R. Marino, L. Aldrighetti, C. Doria, J. McMichael, J. J. Fung

Research output: Contribution to journalArticle

17 Citations (Scopus)

Abstract

One objective of liver transplant evaluation is to identify patients that harbor a hepatoma, but standard screening techniques are not sensitive enough. We trained neural network ensembles to predict the presence of hepatoma in patients with cirrhosis, based on information collected at the time of transplant evaluation. Network architecture and training were modified to handle missing observations. Three ensembles were trained: ensemble A using the subset with no missing observations (528 patients); ensemble B using the complete set, which included missing observations (853 patients); and ensemble C using the smaller subset, originally with complete data, but after a fixed number of observations were deleted (i. e., made 'missing'). Ensemble performance on testing sets was very good. The areas under the ROC curves were 0.91, 0.89, and 0.90, for ensembles A, B, and C, respectively. Neural networks can successfully perform this classification task, and strategies can be developed that allow use of incomplete observations.

Original languageEnglish
Pages (from-to)253-258
Number of pages6
JournalMethods of Information in Medicine
Volume34
Issue number3
Publication statusPublished - 1995

Fingerprint

Hepatocellular Carcinoma
Fibrosis
Transplants
ROC Curve
Area Under Curve
Liver

Keywords

  • Diagnosis
  • Hepatocellular Carcinoma
  • Missing Data
  • Neural Networks

ASJC Scopus subject areas

  • Health Informatics
  • Nursing(all)
  • Health Information Management

Cite this

Building clinical classifiers using incomplete observations - A neural network ensemble for hepatoma detection in patients with cirrhosis. / Doyle, H. R.; Parmanto, B.; Munro, P. W.; Marino, I. R.; Aldrighetti, L.; Doria, C.; McMichael, J.; Fung, J. J.

In: Methods of Information in Medicine, Vol. 34, No. 3, 1995, p. 253-258.

Research output: Contribution to journalArticle

Doyle, H. R. ; Parmanto, B. ; Munro, P. W. ; Marino, I. R. ; Aldrighetti, L. ; Doria, C. ; McMichael, J. ; Fung, J. J. / Building clinical classifiers using incomplete observations - A neural network ensemble for hepatoma detection in patients with cirrhosis. In: Methods of Information in Medicine. 1995 ; Vol. 34, No. 3. pp. 253-258.
@article{b37b6fa586b844159cf8f0ac115b8078,
title = "Building clinical classifiers using incomplete observations - A neural network ensemble for hepatoma detection in patients with cirrhosis",
abstract = "One objective of liver transplant evaluation is to identify patients that harbor a hepatoma, but standard screening techniques are not sensitive enough. We trained neural network ensembles to predict the presence of hepatoma in patients with cirrhosis, based on information collected at the time of transplant evaluation. Network architecture and training were modified to handle missing observations. Three ensembles were trained: ensemble A using the subset with no missing observations (528 patients); ensemble B using the complete set, which included missing observations (853 patients); and ensemble C using the smaller subset, originally with complete data, but after a fixed number of observations were deleted (i. e., made 'missing'). Ensemble performance on testing sets was very good. The areas under the ROC curves were 0.91, 0.89, and 0.90, for ensembles A, B, and C, respectively. Neural networks can successfully perform this classification task, and strategies can be developed that allow use of incomplete observations.",
keywords = "Diagnosis, Hepatocellular Carcinoma, Missing Data, Neural Networks",
author = "Doyle, {H. R.} and B. Parmanto and Munro, {P. W.} and Marino, {I. R.} and L. Aldrighetti and C. Doria and J. McMichael and Fung, {J. J.}",
year = "1995",
language = "English",
volume = "34",
pages = "253--258",
journal = "Methods of Information in Medicine",
issn = "0026-1270",
publisher = "Schattauer GmbH",
number = "3",

}

TY - JOUR

T1 - Building clinical classifiers using incomplete observations - A neural network ensemble for hepatoma detection in patients with cirrhosis

AU - Doyle, H. R.

AU - Parmanto, B.

AU - Munro, P. W.

AU - Marino, I. R.

AU - Aldrighetti, L.

AU - Doria, C.

AU - McMichael, J.

AU - Fung, J. J.

PY - 1995

Y1 - 1995

N2 - One objective of liver transplant evaluation is to identify patients that harbor a hepatoma, but standard screening techniques are not sensitive enough. We trained neural network ensembles to predict the presence of hepatoma in patients with cirrhosis, based on information collected at the time of transplant evaluation. Network architecture and training were modified to handle missing observations. Three ensembles were trained: ensemble A using the subset with no missing observations (528 patients); ensemble B using the complete set, which included missing observations (853 patients); and ensemble C using the smaller subset, originally with complete data, but after a fixed number of observations were deleted (i. e., made 'missing'). Ensemble performance on testing sets was very good. The areas under the ROC curves were 0.91, 0.89, and 0.90, for ensembles A, B, and C, respectively. Neural networks can successfully perform this classification task, and strategies can be developed that allow use of incomplete observations.

AB - One objective of liver transplant evaluation is to identify patients that harbor a hepatoma, but standard screening techniques are not sensitive enough. We trained neural network ensembles to predict the presence of hepatoma in patients with cirrhosis, based on information collected at the time of transplant evaluation. Network architecture and training were modified to handle missing observations. Three ensembles were trained: ensemble A using the subset with no missing observations (528 patients); ensemble B using the complete set, which included missing observations (853 patients); and ensemble C using the smaller subset, originally with complete data, but after a fixed number of observations were deleted (i. e., made 'missing'). Ensemble performance on testing sets was very good. The areas under the ROC curves were 0.91, 0.89, and 0.90, for ensembles A, B, and C, respectively. Neural networks can successfully perform this classification task, and strategies can be developed that allow use of incomplete observations.

KW - Diagnosis

KW - Hepatocellular Carcinoma

KW - Missing Data

KW - Neural Networks

UR - http://www.scopus.com/inward/record.url?scp=0029054260&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029054260&partnerID=8YFLogxK

M3 - Article

C2 - 7666803

AN - SCOPUS:0029054260

VL - 34

SP - 253

EP - 258

JO - Methods of Information in Medicine

JF - Methods of Information in Medicine

SN - 0026-1270

IS - 3

ER -