Derivation and Validation of a Bayesian Network to Predict Pretest Probability of Venous Thromboembolism

Created by W.Langdon from gp-bibliography.bib Revision:1.3973

@Article{Kline:2005:AEM,
  author =       "Jeffrey A. Kline and Andrew J. Novobilski and 
                 Christopher Kabrhel and Peter B. Richman and 
                 D. Mark Courtney",
  title =        "Derivation and Validation of a Bayesian Network to
                 Predict Pretest Probability of Venous Thromboembolism",
  journal =      "Annals of Emergency Medicine",
  year =         "2005",
  volume =       "45",
  number =       "3",
  pages =        "282--290",
  month =        mar,
  keywords =     "genetic algorithms, genetic programming, Bayesian
                 Networks, datamining",
  DOI =          "doi:10.1016/j.annemergmed.2004.08.036",
  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
                 2percent.

                 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 standardised
                 testing, including pulmonary vascular imaging, and
                 90-day follow-up (11.0percent 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
                 (8.0percent were venous thromboembolism positive). The
                 Bayesian network probability estimate was normalised to
                 a score of 0percent to 100percent. Results

                 Of 1,423 patients in the validation cohort, 711
                 (50percent; 95percent confidence interval [CI]
                 47percent to 52percent) had a score less than or equal
                 to 2percent that predicted a low pretest probability.
                 Of these 711 patients, 700 (98.5percent; 95percent CI
                 97.2percent to 99.2percent) 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 (le
                 2percent); 98.5percent of these patients were correctly
                 classified by the network.",
  notes =        "SEE EDITORIAL, P. 291.",
}

Genetic Programming entries for Jeffrey A Kline Andrew J Novobilski Christopher Kabrhel Peter B Richman D Mark Courtney

Citations