Probabilistic asthma case finding

a noisy or reformulation.

Vibha Anand, Stephen Downs

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

Bayesian Networks are used to model domain knowledge with natural perception of causal influences. Even though Bayesian Networks reduce the number of probabilities required to specify relationships in the domain, specifying these probabilities for large networks can be prohibitive. The Noisy-OR formalism of Bayesian Network (BN) overcomes this shortcoming by making an assumption of causal independence among the modeled causes and their common effect. However, the accuracy of this assumption has rarely been tested. In this paper we report the results of an empirical study in the domain of asthma case finding that compares the Noisy-OR reformulation of the expert BN with the expert BN trained using large clinical data set from the Regenstrief Medical Record System. Our results show that the BN with Noisy-OR formulation for this domain performs comparably with the experts BN suggesting that this formalism is robust, at least in this domain.

Original languageEnglish
Pages (from-to)6-10
Number of pages5
JournalAMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium
StatePublished - 2008

Fingerprint

Asthma
Medical Records
Datasets

ASJC Scopus subject areas

  • Medicine(all)

Cite this

@article{52363873eb5f4b1eb02463df5edc8201,
title = "Probabilistic asthma case finding: a noisy or reformulation.",
abstract = "Bayesian Networks are used to model domain knowledge with natural perception of causal influences. Even though Bayesian Networks reduce the number of probabilities required to specify relationships in the domain, specifying these probabilities for large networks can be prohibitive. The Noisy-OR formalism of Bayesian Network (BN) overcomes this shortcoming by making an assumption of causal independence among the modeled causes and their common effect. However, the accuracy of this assumption has rarely been tested. In this paper we report the results of an empirical study in the domain of asthma case finding that compares the Noisy-OR reformulation of the expert BN with the expert BN trained using large clinical data set from the Regenstrief Medical Record System. Our results show that the BN with Noisy-OR formulation for this domain performs comparably with the experts BN suggesting that this formalism is robust, at least in this domain.",
author = "Vibha Anand and Stephen Downs",
year = "2008",
language = "English",
pages = "6--10",
journal = "AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium",
issn = "1559-4076",
publisher = "American Medical Informatics Association",

}

TY - JOUR

T1 - Probabilistic asthma case finding

T2 - a noisy or reformulation.

AU - Anand, Vibha

AU - Downs, Stephen

PY - 2008

Y1 - 2008

N2 - Bayesian Networks are used to model domain knowledge with natural perception of causal influences. Even though Bayesian Networks reduce the number of probabilities required to specify relationships in the domain, specifying these probabilities for large networks can be prohibitive. The Noisy-OR formalism of Bayesian Network (BN) overcomes this shortcoming by making an assumption of causal independence among the modeled causes and their common effect. However, the accuracy of this assumption has rarely been tested. In this paper we report the results of an empirical study in the domain of asthma case finding that compares the Noisy-OR reformulation of the expert BN with the expert BN trained using large clinical data set from the Regenstrief Medical Record System. Our results show that the BN with Noisy-OR formulation for this domain performs comparably with the experts BN suggesting that this formalism is robust, at least in this domain.

AB - Bayesian Networks are used to model domain knowledge with natural perception of causal influences. Even though Bayesian Networks reduce the number of probabilities required to specify relationships in the domain, specifying these probabilities for large networks can be prohibitive. The Noisy-OR formalism of Bayesian Network (BN) overcomes this shortcoming by making an assumption of causal independence among the modeled causes and their common effect. However, the accuracy of this assumption has rarely been tested. In this paper we report the results of an empirical study in the domain of asthma case finding that compares the Noisy-OR reformulation of the expert BN with the expert BN trained using large clinical data set from the Regenstrief Medical Record System. Our results show that the BN with Noisy-OR formulation for this domain performs comparably with the experts BN suggesting that this formalism is robust, at least in this domain.

UR - http://www.scopus.com/inward/record.url?scp=73949085811&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=73949085811&partnerID=8YFLogxK

M3 - Article

SP - 6

EP - 10

JO - AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium

JF - AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium

SN - 1559-4076

ER -