Neural Network Predictions of Significant Coronary Artery Stenosis in Women

  • B. A. Mobley
  • W. E. Moore
  • E. Schechter
  • J. E. Eichner
  • P. A. McKee
Part of the Studies in Fuzziness and Soft Computing book series (STUDFUZZ, volume 96)


Records from 1009 female coronary angiography patients were analyzed by artificial neural network. An outcome in each record, any coronary artery stenosis greater than 50% formed the dichotomous supervisory variable for the neural network. The network contained 19 and 30 elements in the input and middle layers respectively. A single output element corresponded to the supervisory variable. Patient records were ordered according to the date of the angiography and placed into four files. The first 409 records comprised the training file on which the network was trained by the method of backpropagation of errors with momentum. The next 400 records formed a cross-validation file on which the performance of the trained network was optimized. The next 100 records was a cutoff determination file, the file used to determine the output of the network, which distinguished significant stenosis. The cutoff was applied to the 100 records of the test file. ROC analysis revealed that a cutoff of 0.30 maximized specificity while maintaining perfect sensitivity in the cutoff determination file. The cutoff of 0.30 also maintained perfect sensitivity in the test file, while the trained network made output predictions less than 0.30 for 9 of 38 (24%) women who had no stenosis greater than 50%. Therefore the neural network system allowed the correct identification of 24% of the false positives from pre-angiographic diagnostic systems without making a single false negative prediction.


Receiver Operating Characteristic Artificial Neural Network Coronary Angiography Receiver Operating Characteristic Curve Receiver Operating Characteristic Analysis 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Graboys, T.B., Biegelsen, B., Lampert, S., Blatt, C.M., and Lown, B. (1992), “Results of a second-opinion trial among patients recommended for coronary angiography,” JAMA, vol. 268, pp. 2537–2540.PubMedCrossRefGoogle Scholar
  2. [2]
    Ouzan, J., Chapoutot, L., Carre, E.J., Liehn, J.C., and Elaerts, J. (1993), “Multivariate analysis of the diagnostic values of clinical examination, exercise testing and exercise radionuclide angiography in coronary artery disease,” Cardiology, vol. 83, pp. 197–204.Google Scholar
  3. [3]
    Guerci, A.D., Spadaro, L.A., Goodman, K.J., Lledo-Perez, A., Newstein, D., Lerner, G., and Arad, Y. (1998), “Comparison of electron beam computed tomography scanning and conventional risk factor assessment for the prediction of angiographic coronary artery disease,” JACC, vol. 32, pp. 673–679.PubMedCrossRefGoogle Scholar
  4. [4]
    Mobley, B.A., Schechter, E., Moore, W.E., McKee, P.A., and Eichner, J.E. (2000), “Predictions of coronary artery stenosis by artificial neural network,” Art. Intell. Med., vol. 18, pp. 187–203.CrossRefGoogle Scholar
  5. [5]
    Ayanian, J.Z. and Epstein, A.M., (1991), “Differences in the use of procedures between women and men hospitalized for coronary heart disease,” N. Engl. J Med., vol. 325, pp. 221–225.PubMedCrossRefGoogle Scholar
  6. [6]
    Jong, P., Mohammed, S., and Sternberg, L., (1996), “Sex differences in the features of coronary artery disease of patients undergoing coronary angiography,” Can. J Cardiol., vol. 12, pp. 671–677.PubMedGoogle Scholar
  7. [7]
    Baxt, W.G., (1991), “Use of an artificial neural network for the diagnosis of myocardial infarction,” Ann. Int. Med., vol. 115, pp. 843–848.PubMedCrossRefGoogle Scholar
  8. [8]
    Baxt, W.G. and Skora, J. (1996), “Prospective validation of artificial neural network trained to identify acute myocardial infarction,” Lancet, vol. 347, pp. 12–15.PubMedCrossRefGoogle Scholar
  9. [9]
    Wasserman, P.D. (1989), Neural Computing: Theory and Practice, Van Nostrand, Reinhold, New York.Google Scholar
  10. [10]
    Caudill, M. and Butler, C. (1990), Naturally Intelligent Systems, MIT Press, Cambridge, MA.Google Scholar
  11. [11]
    Hertz, J., Krogh, A., and Palmer, R.G. (1991), Introduction to the Theory of Neural Computation, Addison-Wesley, Redwood City, CA.Google Scholar
  12. [12]
    Rumelhart, D.E., Hinton, G.E., and Williams, R.J., (1986), “Learning internal representations by error propagation,” Parallel Distributed Processing, pp. 318–362.Google Scholar
  13. [13]
    Hosmer, D.W. and Lemeshow, S. (1989), Applied Logistic Regression, John Wiley & Sons, New YorkGoogle Scholar
  14. [14]
    Kleinbaum, D.G. (1994), Logistic Regression: a Self-Learning Text, Springer-Verlag, New YorkGoogle Scholar
  15. [15]
    Kleinbaum, D.G., Kupper, L.L., and Muller, K.E. (1988), Applied Regression Analysis and Other Multivariable Methods, Duxbury Press, Belmont, CA.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2002

Authors and Affiliations

  • B. A. Mobley
  • W. E. Moore
  • E. Schechter
  • J. E. Eichner
  • P. A. McKee

There are no affiliations available

Personalised recommendations