Back

An interpretable and explainable neural network to classify sports-related cardiac arrhythmias in professional football athletes

Vanegas Mueller, E.; Harford, M.; He, L.; Banerjee, A.; Leeson, P.; Villarroel, M.

2026-03-02 sports medicine
10.64898/2026.02.23.26346628 medRxiv
Show abstract

Sudden cardiac death risk is 2-3-fold higher in athletes than in non-athletes. We classify sports-related cardiac arrhythmias using a novel explainability framework comprising data analysis, model interpretability, post-hoc visualisation, and systematic assessment. Two neural networks--one with interpretable sinc convolution and one with standard convolution--were trained on general-population ECGs (PhysioNet, n=88,253, 30 arrhythmias, three continents) and tested on professional footballers (PF12RED, n=161) via domain adaptation for normal sinus rhythm (NSR), sinus bradycardia (SB), incomplete right bundle branch block (IRBBB), and T-wave inversion (TWI). Sinc convolution achieved superior NSR detection (AUROC 0.75 vs 0.70), whilst standard convolution excelled at SB (0.74 vs 0.73), IRBBB (0.66 vs 0.58), and TWI (0.59 vs 0.54). Gradient-weighted Class Activation Mapping revealed that sinc models focus on physiologically relevant ECG segments (the PR interval for NSR/SB and the T wave for TWI). We hypothesise that sinc convolution better captures periodic rhythms but struggles with complex morphological patterns, suggesting architectural choice should align with underlying cardiac pathophysiology. Graphical abstractAbbreviations: AI, artificial intelligence; AUPRC, area under the precision-recall curve; AUROC, area under the receiver operating characteristic curve; Conv, convolution; ECG, electrocardiogram; Grad-CAM, gradient-weighted class activation mapping; IAVB, first-degree atrioventricular block; IRBBB, incomplete right bundle branch block; LAD, left axis deviation; LBBB, left bundle branch block; LVH, left ventricular hypertrophy; NSR, normal sinus rhythm; QT, QT interval; RAD, right axis deviation; RBBB, right bundle branch block; RVH, right ventricular hypertrophy; SA, sinus arrhythmia; SB, sinus bradycardia; TWI, T-wave inversion; xAI, explainable artificial intelligence. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=123 SRC="FIGDIR/small/26346628v1_ufig1.gif" ALT="Figure 1"> View larger version (60K): org.highwire.dtl.DTLVardef@124f232org.highwire.dtl.DTLVardef@98ba6forg.highwire.dtl.DTLVardef@f7fdedorg.highwire.dtl.DTLVardef@14012b3_HPS_FORMAT_FIGEXP M_FIG C_FIG

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

1
Communications Biology
886 papers in training set
Top 0.1%
10.0%
2
Frontiers in Cardiovascular Medicine
49 papers in training set
Top 0.3%
10.0%
3
PLOS Computational Biology
1633 papers in training set
Top 3%
10.0%
4
Frontiers in Genetics
197 papers in training set
Top 0.5%
8.3%
5
Scientific Reports
3102 papers in training set
Top 15%
6.7%
6
PLOS ONE
4510 papers in training set
Top 29%
6.2%
50% of probability mass above
7
Computers in Biology and Medicine
120 papers in training set
Top 0.8%
3.6%
8
European Heart Journal - Digital Health
15 papers in training set
Top 0.2%
3.6%
9
Journal of the American Heart Association
119 papers in training set
Top 2%
3.5%
10
Frontiers in Physiology
93 papers in training set
Top 2%
2.6%
11
Physiological Measurement
12 papers in training set
Top 0.2%
1.9%
12
npj Digital Medicine
97 papers in training set
Top 2%
1.9%
13
Cell Reports
1338 papers in training set
Top 23%
1.8%
14
iScience
1063 papers in training set
Top 15%
1.7%
15
Circulation
66 papers in training set
Top 2%
1.6%
16
Biology Methods and Protocols
53 papers in training set
Top 1%
1.6%
17
Patterns
70 papers in training set
Top 1%
1.5%
18
Computer Methods and Programs in Biomedicine
27 papers in training set
Top 0.4%
1.5%
19
Computational and Structural Biotechnology Journal
216 papers in training set
Top 6%
1.3%
20
Medical Image Analysis
33 papers in training set
Top 0.7%
1.3%
21
PLOS Digital Health
91 papers in training set
Top 2%
0.9%
22
PLOS Biology
408 papers in training set
Top 18%
0.8%
23
Journal of Neural Engineering
197 papers in training set
Top 2%
0.8%
24
European Journal of Epidemiology
40 papers in training set
Top 0.8%
0.7%
25
Bioinformatics
1061 papers in training set
Top 10%
0.6%
26
Journal of the American Medical Informatics Association
61 papers in training set
Top 2%
0.6%
27
PLOS Neglected Tropical Diseases
378 papers in training set
Top 6%
0.6%