Derivation and validation of a bayesian network to predict pretest probability of venous thromboembolism

Jeffrey A. Kline, Andrew J. Novobilski, Christopher Kabrhel, Peter B. Richman, D. Mark Courtney

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

Study objective: A Bayesian network can estimate a numeric pretest probability of venous thromboembolism on the basis of values of clinical variables. We determine the accuracy with which a Bayesian network can identify patients with a low pretest probability of venous thromboembolism, defined as less than or equal to 2%. Methods: Using commercial software, we derived a population of Bayesian networks from 25 input variables collected on 3,145 emergency department (ED) patients with suspected venous thromboembolism who underwent standardized testing, including pulmonary vascular imaging, and 90-day follow-up (11.0% of patients were venous thromboembolism positive). The best-fit Bayesian network was selected using a genetic algorithm. The selected Bayesian network was tested in a validation population of 1,423 ED patients prospectively evaluated for venous thromboembolism, including 90-day follow-up (8.0% were venous thromboembolism positive). The Bayesian network probability estimate was normalized to a score of 0% to 100%. Results: Of 1,423 patients in the validation cohort, 711 (50%; 95% confidence interval [CI] 47% to 52%) had a score less than or equal to 2% that predicted a low pretest probability. Of these 711 patients, 700 (98.5%; 95% CI 97.2% to 99.2%) had no venous thromboembolism at follow-up. Conclusion: A Bayesian network, derived and independently validated in ED populations, identified half of the validation cohort as having a low pretest probability (≤2%); 98.5% of these patients were correctly classified by the network.

Original languageEnglish (US)
Pages (from-to)282-290
Number of pages9
JournalAnnals of Emergency Medicine
Volume45
Issue number3
DOIs
StatePublished - Mar 2005
Externally publishedYes

Fingerprint

Venous Thromboembolism
Hospital Emergency Service
Confidence Intervals
Population
Blood Vessels
Software
Lung

ASJC Scopus subject areas

  • Emergency Medicine

Cite this

Derivation and validation of a bayesian network to predict pretest probability of venous thromboembolism. / Kline, Jeffrey A.; Novobilski, Andrew J.; Kabrhel, Christopher; Richman, Peter B.; Courtney, D. Mark.

In: Annals of Emergency Medicine, Vol. 45, No. 3, 03.2005, p. 282-290.

Research output: Contribution to journalArticle

Kline, Jeffrey A. ; Novobilski, Andrew J. ; Kabrhel, Christopher ; Richman, Peter B. ; Courtney, D. Mark. / Derivation and validation of a bayesian network to predict pretest probability of venous thromboembolism. In: Annals of Emergency Medicine. 2005 ; Vol. 45, No. 3. pp. 282-290.
@article{2c44759b0c58484e8f005461793628cf,
title = "Derivation and validation of a bayesian network to predict pretest probability of venous thromboembolism",
abstract = "Study objective: A Bayesian network can estimate a numeric pretest probability of venous thromboembolism on the basis of values of clinical variables. We determine the accuracy with which a Bayesian network can identify patients with a low pretest probability of venous thromboembolism, defined as less than or equal to 2{\%}. Methods: Using commercial software, we derived a population of Bayesian networks from 25 input variables collected on 3,145 emergency department (ED) patients with suspected venous thromboembolism who underwent standardized testing, including pulmonary vascular imaging, and 90-day follow-up (11.0{\%} of patients were venous thromboembolism positive). The best-fit Bayesian network was selected using a genetic algorithm. The selected Bayesian network was tested in a validation population of 1,423 ED patients prospectively evaluated for venous thromboembolism, including 90-day follow-up (8.0{\%} were venous thromboembolism positive). The Bayesian network probability estimate was normalized to a score of 0{\%} to 100{\%}. Results: Of 1,423 patients in the validation cohort, 711 (50{\%}; 95{\%} confidence interval [CI] 47{\%} to 52{\%}) had a score less than or equal to 2{\%} that predicted a low pretest probability. Of these 711 patients, 700 (98.5{\%}; 95{\%} CI 97.2{\%} to 99.2{\%}) had no venous thromboembolism at follow-up. Conclusion: A Bayesian network, derived and independently validated in ED populations, identified half of the validation cohort as having a low pretest probability (≤2{\%}); 98.5{\%} of these patients were correctly classified by the network.",
author = "Kline, {Jeffrey A.} and Novobilski, {Andrew J.} and Christopher Kabrhel and Richman, {Peter B.} and Courtney, {D. Mark}",
year = "2005",
month = "3",
doi = "10.1016/j.annemergmed.2004.08.036",
language = "English (US)",
volume = "45",
pages = "282--290",
journal = "Annals of Emergency Medicine",
issn = "0196-0644",
publisher = "Mosby Inc.",
number = "3",

}

TY - JOUR

T1 - Derivation and validation of a bayesian network to predict pretest probability of venous thromboembolism

AU - Kline, Jeffrey A.

AU - Novobilski, Andrew J.

AU - Kabrhel, Christopher

AU - Richman, Peter B.

AU - Courtney, D. Mark

PY - 2005/3

Y1 - 2005/3

N2 - Study objective: A Bayesian network can estimate a numeric pretest probability of venous thromboembolism on the basis of values of clinical variables. We determine the accuracy with which a Bayesian network can identify patients with a low pretest probability of venous thromboembolism, defined as less than or equal to 2%. Methods: Using commercial software, we derived a population of Bayesian networks from 25 input variables collected on 3,145 emergency department (ED) patients with suspected venous thromboembolism who underwent standardized testing, including pulmonary vascular imaging, and 90-day follow-up (11.0% of patients were venous thromboembolism positive). The best-fit Bayesian network was selected using a genetic algorithm. The selected Bayesian network was tested in a validation population of 1,423 ED patients prospectively evaluated for venous thromboembolism, including 90-day follow-up (8.0% were venous thromboembolism positive). The Bayesian network probability estimate was normalized to a score of 0% to 100%. Results: Of 1,423 patients in the validation cohort, 711 (50%; 95% confidence interval [CI] 47% to 52%) had a score less than or equal to 2% that predicted a low pretest probability. Of these 711 patients, 700 (98.5%; 95% CI 97.2% to 99.2%) had no venous thromboembolism at follow-up. Conclusion: A Bayesian network, derived and independently validated in ED populations, identified half of the validation cohort as having a low pretest probability (≤2%); 98.5% of these patients were correctly classified by the network.

AB - Study objective: A Bayesian network can estimate a numeric pretest probability of venous thromboembolism on the basis of values of clinical variables. We determine the accuracy with which a Bayesian network can identify patients with a low pretest probability of venous thromboembolism, defined as less than or equal to 2%. Methods: Using commercial software, we derived a population of Bayesian networks from 25 input variables collected on 3,145 emergency department (ED) patients with suspected venous thromboembolism who underwent standardized testing, including pulmonary vascular imaging, and 90-day follow-up (11.0% of patients were venous thromboembolism positive). The best-fit Bayesian network was selected using a genetic algorithm. The selected Bayesian network was tested in a validation population of 1,423 ED patients prospectively evaluated for venous thromboembolism, including 90-day follow-up (8.0% were venous thromboembolism positive). The Bayesian network probability estimate was normalized to a score of 0% to 100%. Results: Of 1,423 patients in the validation cohort, 711 (50%; 95% confidence interval [CI] 47% to 52%) had a score less than or equal to 2% that predicted a low pretest probability. Of these 711 patients, 700 (98.5%; 95% CI 97.2% to 99.2%) had no venous thromboembolism at follow-up. Conclusion: A Bayesian network, derived and independently validated in ED populations, identified half of the validation cohort as having a low pretest probability (≤2%); 98.5% of these patients were correctly classified by the network.

UR - http://www.scopus.com/inward/record.url?scp=13844311801&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=13844311801&partnerID=8YFLogxK

U2 - 10.1016/j.annemergmed.2004.08.036

DO - 10.1016/j.annemergmed.2004.08.036

M3 - Article

C2 - 15726051

AN - SCOPUS:13844311801

VL - 45

SP - 282

EP - 290

JO - Annals of Emergency Medicine

JF - Annals of Emergency Medicine

SN - 0196-0644

IS - 3

ER -