Skip to main content
Springer Nature Link
Log in

A Comparison of Linear and Non-linear Classifiers for the Detection of Coronary Artery Disease in Stress-ECG

  • Conference paper
  • First Online:
Artificial Intelligence in Medicine (AIMDM 1999)

Part of the book series: Lecture Notes in Computer Science ((LNAI,volume 1620))

Abstract

In this paper we report about a retrospective comparative study on three classifiers (multilayer perceptron, logistic classifier, and nearest neighbor classifier) applied to the task of detecting coronary artery disease in variables obtained from stress-ECG (treadmill exercise). A 10-fold cross-validation on all three methods was applied and the results were compared to expert performance. The results indicate that the multilayer perceptron had significantly higher specificity (correctly classified normals) than both the other classifiers and experts. In addition, they perform with lower standard deviation than experts, pointing to a more reliable, objective measure for diagnosis.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution

Institutional subscriptions

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Similar content being viewed by others

References

  1. Detrano R., et al.: Exercise-induced ST segment depression in the diagnosis of multivessel coronary disease: a meta analysis. J. Am. Coll. Cardiol. 14(1989) 1501–1508

    Article  Google Scholar 

  2. Hurst W.: The Heart, 7th ed., McGraw Hill, New York (1990)

    Google Scholar 

  3. Julian D.G., Camm A.J., Fox K.M., Hall R.J.C., Poole-Wilson P.A.: Diseases of the Heart. Bailliere Tindall, London (1989)

    Google Scholar 

  4. Koller H., Leitgeb E.: Ist eine Verbesserung der Aussagekraft und Reproduzier-barkeit des Belastungs-EKG erzielbar?. Wiener Medizinische Wochenschrift 143 (1993) 110–117

    Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Austrian Research Institute for Artificial Intelligence, Schottengasse 3, A-1010, Vienna, Austria

    Georg Dorffner

  2. Dept. of Medical Cybernetics and AI, Univ. of Vienna, Vienna, Austria

    Georg Dorffner, Ernst Leitgeb & Heinz Koller

  3. Leitgeb Company, Strasshof, Austria

    Ernst Leitgeb

  4. LKH, Klagenfurt, Austria

    Heinz Koller

Authors
  1. Georg Dorffner
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Ernst Leitgeb
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Heinz Koller
    View author publications

    You can also search for this author in PubMed Google Scholar

Editor information

Editors and Affiliations

  1. Austrian Research Institute for Artificial Intelligence, Schottengasse 3, A-1010, Vienna, Austria

    Werner Horn

  2. Department of Medicine and Computer Science, Stanford University, 251 Campus Drive, Stanford, CA, 94305-5479, USA

    Yuval Shahar

  3. Department of Medicine Karolinska Institutet, Huddinge University Hospital, Sweden

    Greger Lindberg

  4. Department of Medical Informatics and Image Analysis, Aalborg University, Fredrik Bajers Vej 7D, DK-9000, Aalborg, Denmark

    Steen Andreassen

  5. School of Public Policy, University College London, 29/30 Tavistock Square, London, WC1H 9EZ, UK

    Jeremy Wyatt

Rights and permissions

Reprints and permissions

Copyright information

© 1999 Springer-Verlag Berlin Heidelberg

About this paper

Cite this paper

Dorffner, G., Leitgeb, E., Koller, H. (1999). A Comparison of Linear and Non-linear Classifiers for the Detection of Coronary Artery Disease in Stress-ECG. In: Horn, W., Shahar, Y., Lindberg, G., Andreassen, S., Wyatt, J. (eds) Artificial Intelligence in Medicine. AIMDM 1999. Lecture Notes in Computer Science(), vol 1620. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-48720-4_24

Download citation

  • DOI: https://doi.org/10.1007/3-540-48720-4_24

  • Published:

  • Publisher Name: Springer, Berlin, Heidelberg

  • Print ISBN: 978-3-540-66162-7

  • Online ISBN: 978-3-540-48720-3

  • eBook Packages: Springer Book Archive

Publish with us

Policies and ethics

Search

Navigation