Table of Contents Author Guidelines Submit a Manuscript
BioMed Research International
Volume 2015 (2015), Article ID 680381, 7 pages
Research Article

Disease Classification and Biomarker Discovery Using ECG Data

Department of Statistics and Actuarial Sciences, East China Normal University, Shanghai 200241, China

Received 26 August 2015; Revised 2 November 2015; Accepted 10 November 2015

Academic Editor: Cristiana Corsi

Copyright © 2015 Rong Huang and Yingchun Zhou. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


In the recent decade, disease classification and biomarker discovery have become increasingly important in modern biological and medical research. ECGs are comparatively low-cost and noninvasive in screening and diagnosing heart diseases. With the development of personal ECG monitors, large amounts of ECGs are recorded and stored; therefore, fast and efficient algorithms are called for to analyze the data and make diagnosis. In this paper, an efficient and easy-to-interpret procedure of cardiac disease classification is developed through novel feature extraction methods and comparison of classifiers. Motivated by the observation that the distributions of various measures on ECGs of the diseased group are often skewed, heavy-tailed, or multimodal, we characterize the distributions by sample quantiles which outperform sample means. Three classifiers are compared in application both to all features and to dimension-reduced features by PCA: stepwise discriminant analysis (SDA), SVM, and LASSO logistic regression. It is found that SDA applied to dimension-reduced features by PCA is the most stable and effective procedure, with sensitivity, specificity, and accuracy being 89.68%, 84.62%, and 88.52%, respectively.