Diagnosis of airway obstruction or restrictive spirometric patterns by multiclass support vector machines

J Med Syst. 2010 Oct;34(5):967-73. doi: 10.1007/s10916-009-9312-7. Epub 2009 May 12.

Abstract

This paper presents the use of multiclass support vector machines (SVMs) for diagnosis of spirometric patterns (normal, restrictive, obstructive). The SVM decisions were fused using the error correcting output codes (ECOC). The multiclass SVM with the ECOC was trained on three spirometric parameters (forced expiratory volume in 1s--FEV1, forced vital capacity--FVC and FEV1/FVC ratio). The total classification accuracy of the SVM is 97.32%. The obtained results confirmed the validity of the SVMs to help in clinical decision-making.

Publication types

  • Clinical Trial

MeSH terms

  • Adult
  • Forced Expiratory Volume
  • Humans
  • Male
  • Neural Networks, Computer*
  • Pulmonary Disease, Chronic Obstructive / diagnosis*
  • Respiratory Insufficiency / diagnosis*
  • Sensitivity and Specificity
  • Spirometry / instrumentation*
  • Vital Capacity