Mehari, T; Sundar, A; Bosnjakovic, A; Harris, P M; Williams, S E; Loewe, A; Doessel, O; Nagel, C; Strodthoff, N; Aston, P J (2024) ECG Feature Importance Rankings: Cardiologists vs. Algorithms. IEEE Journal of Biomedical and Health Informatics, 28 (4). pp. 2014-2024.

Preview

Text eid10007.pdf - Published Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (2MB) | Preview

Abstract

Feature importance methods promise to provide a ranking of features according to importance for a given classification task. A wide range of methods exist but their rankings often disagree and they are inherently difficult to evaluate due to a lack of ground truth beyond synthetic datasets. In this work, we put feature importance methods to the test on real-world data in the domain of cardiology, where we try to distinguish three specific pathologies from healthy subjects based on ECG features comparing to features used in cardiologists' decision rules as ground truth. Some methods generally performed well and others performed poorly, while some methods did well on some but not all of the problems considered.

Item Type:	Article
Keywords:	Electrocardiogram, feature importance ranking, cardiologist, atrioventricular block, right branch bundle block, left branch bundle block
Subjects:	Mathematics and Scientific Computing > Signal Processing
Divisions:	Data Science
Identification number/DOI:	10.1109/JBHI.2024.3354301
Last Modified:	10 Sep 2024 10:35
URI:	https://eprintspublications.npl.co.uk/id/eprint/10007