Prediction of community-acquired pneumonia using artificial neural networks

Paul S. Heckerling, Ben S. Gerber, Thomas Gerald Tape, Robert Swift Wigton

Research output: Contribution to journalArticle

14 Citations (Scopus)

Abstract

Background. Artificial neural networks (ANN) have been used in the prediction of several medical conditions but have not been previously used to predict pneumonia. The authors used ANN to predict the presence or absence of pneumonia among patients presenting to the emergency department with acute respiratory complaints and compared the results with those obtained using logistic regression modeling. Methods. Feed-forward back-propagation ANN were trained on sociodemographic, symptom, sign, comorbidity, and radiographic outcome data among 1044 patients from the University of Illinois (the training cohort) and were applied to 116 patients from the University of Nebraska (the testing cohort). ANN trained using different strategies were compared to each other and to main-effects logistic regression. Calibration accuracy was measured as mean square error and discrimination accuracy as the area under a receiver operating characteristic (ROC) curve. Results. A 1 hidden-layer ANN trained using oversampling of pneumonia cases had an ROC area in the training cohort of 0.895, which was greater than the area of 0.840 for logistic regression (P = 0.026). This ANN had an ROC area in the testing cohort of 0.872, not significantly different from its area in the training cohort (P = 0.597). Operating at a threshold of 0.25, the ANN would have detected 94% to 95% of patients with pneumonia in the 2 cohorts while correctly excluding 39% to 50% of patients with other conditions. ANN trained using other strategies discriminated equally in the 2 cohorts but no better than did logistic regression. Conclusions. Among adults presenting with acute respiratory illness, ANN accurately discriminated patients with and without pneumonia and, under some circumstances, improved on the accuracy of logistic regression.

Original languageEnglish (US)
Pages (from-to)112-121
Number of pages10
JournalMedical Decision Making
Volume23
Issue number2
DOIs
StatePublished - Mar 1 2003

Fingerprint

Pneumonia
Logistic Models
ROC Curve
Calibration
Signs and Symptoms
Hospital Emergency Service
Comorbidity

Keywords

  • Artificial neural networks
  • Logistic regression
  • Pneumonia

ASJC Scopus subject areas

  • Health Policy

Cite this

Prediction of community-acquired pneumonia using artificial neural networks. / Heckerling, Paul S.; Gerber, Ben S.; Tape, Thomas Gerald; Wigton, Robert Swift.

In: Medical Decision Making, Vol. 23, No. 2, 01.03.2003, p. 112-121.

Research output: Contribution to journalArticle

@article{2cd2106d40464a91b81886c7535fa342,
title = "Prediction of community-acquired pneumonia using artificial neural networks",
abstract = "Background. Artificial neural networks (ANN) have been used in the prediction of several medical conditions but have not been previously used to predict pneumonia. The authors used ANN to predict the presence or absence of pneumonia among patients presenting to the emergency department with acute respiratory complaints and compared the results with those obtained using logistic regression modeling. Methods. Feed-forward back-propagation ANN were trained on sociodemographic, symptom, sign, comorbidity, and radiographic outcome data among 1044 patients from the University of Illinois (the training cohort) and were applied to 116 patients from the University of Nebraska (the testing cohort). ANN trained using different strategies were compared to each other and to main-effects logistic regression. Calibration accuracy was measured as mean square error and discrimination accuracy as the area under a receiver operating characteristic (ROC) curve. Results. A 1 hidden-layer ANN trained using oversampling of pneumonia cases had an ROC area in the training cohort of 0.895, which was greater than the area of 0.840 for logistic regression (P = 0.026). This ANN had an ROC area in the testing cohort of 0.872, not significantly different from its area in the training cohort (P = 0.597). Operating at a threshold of 0.25, the ANN would have detected 94{\%} to 95{\%} of patients with pneumonia in the 2 cohorts while correctly excluding 39{\%} to 50{\%} of patients with other conditions. ANN trained using other strategies discriminated equally in the 2 cohorts but no better than did logistic regression. Conclusions. Among adults presenting with acute respiratory illness, ANN accurately discriminated patients with and without pneumonia and, under some circumstances, improved on the accuracy of logistic regression.",
keywords = "Artificial neural networks, Logistic regression, Pneumonia",
author = "Heckerling, {Paul S.} and Gerber, {Ben S.} and Tape, {Thomas Gerald} and Wigton, {Robert Swift}",
year = "2003",
month = "3",
day = "1",
doi = "10.1177/0272989X03251247",
language = "English (US)",
volume = "23",
pages = "112--121",
journal = "Medical Decision Making",
issn = "0272-989X",
publisher = "SAGE Publications Inc.",
number = "2",

}

TY - JOUR

T1 - Prediction of community-acquired pneumonia using artificial neural networks

AU - Heckerling, Paul S.

AU - Gerber, Ben S.

AU - Tape, Thomas Gerald

AU - Wigton, Robert Swift

PY - 2003/3/1

Y1 - 2003/3/1

N2 - Background. Artificial neural networks (ANN) have been used in the prediction of several medical conditions but have not been previously used to predict pneumonia. The authors used ANN to predict the presence or absence of pneumonia among patients presenting to the emergency department with acute respiratory complaints and compared the results with those obtained using logistic regression modeling. Methods. Feed-forward back-propagation ANN were trained on sociodemographic, symptom, sign, comorbidity, and radiographic outcome data among 1044 patients from the University of Illinois (the training cohort) and were applied to 116 patients from the University of Nebraska (the testing cohort). ANN trained using different strategies were compared to each other and to main-effects logistic regression. Calibration accuracy was measured as mean square error and discrimination accuracy as the area under a receiver operating characteristic (ROC) curve. Results. A 1 hidden-layer ANN trained using oversampling of pneumonia cases had an ROC area in the training cohort of 0.895, which was greater than the area of 0.840 for logistic regression (P = 0.026). This ANN had an ROC area in the testing cohort of 0.872, not significantly different from its area in the training cohort (P = 0.597). Operating at a threshold of 0.25, the ANN would have detected 94% to 95% of patients with pneumonia in the 2 cohorts while correctly excluding 39% to 50% of patients with other conditions. ANN trained using other strategies discriminated equally in the 2 cohorts but no better than did logistic regression. Conclusions. Among adults presenting with acute respiratory illness, ANN accurately discriminated patients with and without pneumonia and, under some circumstances, improved on the accuracy of logistic regression.

AB - Background. Artificial neural networks (ANN) have been used in the prediction of several medical conditions but have not been previously used to predict pneumonia. The authors used ANN to predict the presence or absence of pneumonia among patients presenting to the emergency department with acute respiratory complaints and compared the results with those obtained using logistic regression modeling. Methods. Feed-forward back-propagation ANN were trained on sociodemographic, symptom, sign, comorbidity, and radiographic outcome data among 1044 patients from the University of Illinois (the training cohort) and were applied to 116 patients from the University of Nebraska (the testing cohort). ANN trained using different strategies were compared to each other and to main-effects logistic regression. Calibration accuracy was measured as mean square error and discrimination accuracy as the area under a receiver operating characteristic (ROC) curve. Results. A 1 hidden-layer ANN trained using oversampling of pneumonia cases had an ROC area in the training cohort of 0.895, which was greater than the area of 0.840 for logistic regression (P = 0.026). This ANN had an ROC area in the testing cohort of 0.872, not significantly different from its area in the training cohort (P = 0.597). Operating at a threshold of 0.25, the ANN would have detected 94% to 95% of patients with pneumonia in the 2 cohorts while correctly excluding 39% to 50% of patients with other conditions. ANN trained using other strategies discriminated equally in the 2 cohorts but no better than did logistic regression. Conclusions. Among adults presenting with acute respiratory illness, ANN accurately discriminated patients with and without pneumonia and, under some circumstances, improved on the accuracy of logistic regression.

KW - Artificial neural networks

KW - Logistic regression

KW - Pneumonia

UR - http://www.scopus.com/inward/record.url?scp=0037351361&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0037351361&partnerID=8YFLogxK

U2 - 10.1177/0272989X03251247

DO - 10.1177/0272989X03251247

M3 - Article

VL - 23

SP - 112

EP - 121

JO - Medical Decision Making

JF - Medical Decision Making

SN - 0272-989X

IS - 2

ER -