Document Type
Conference Proceeding
Language
eng
Format of Original
4 p.
Publication Date
9-19-2004
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Source Publication
Computers in Cardiology, 2004
Source ISSN
0-7803-8927-1
Abstract
In this paper, a method is proposed to improve an algorithm for myocardial ischemia classification created by Langley et al. The Langley classifier achieves a very high sensitivity (99.0%), but a lower specificity value (93.3%). In order to improve the specificity, the proposed algorithm attempts to reclassify the events that the Langley classifier labels ischemic. The classifier used is a support vector machine. The features used are the mean of the ST deviation, maximum value of the ST deviation, and the initial ST deviation. The classifier is able to increase the specificity from 92.3% to 93.3%. The drawback is that the sensitivity is reduced from 99.0% to 97.5%. This causes the overall accuracy to decrease slightly from 95.6 to 94.8. The algorithm shows promise in being able to increase specificity, but work must be done to find features that do not cause such a large decrease in the sensitivity.
Recommended Citation
Zimmerman, M. W. and Povinelli, Richard J., "On Improving the Classification of Myocardial Ischemia Using Holter ECG Data" (2004). Electrical and Computer Engineering Faculty Research and Publications. 163.
https://epublications.marquette.edu/electric_fac/163
Comments
Accepted version. Published as part of the proceedings of the conference, Computers in Cardiology, 2004: 377-380. DOI. © 2004 The Institute of Electrical and Electronics Engineers. Used with permission.