Analyzing ECG signals in professional football players using machine learning techniques

Area of Science:

• Sports Medicine
• Cardiology
• Artificial Intelligence in Healthcare

Background:

• Sudden cardiac arrest is a critical concern in football, with professional players exhibiting distinct electrocardiogram (ECG) signals compared to the general population.
• Existing research on athlete ECGs is limited, despite extensive studies on the general population.

Purpose of the Study:

• To establish a novel 12-lead resting ECG database for elite male football players.
• To develop a free software tool for ECG visualization, denoising, filtering, and automated wave labeling.
• To utilize machine learning (ML) for analyzing ECG wave shapes and automating arrhythmia diagnosis.

Main Methods:

• A prospective observational cohort study collected 10-second, 12-lead ECGs from 54 La Liga professional football players.
• The Pro Football 12-lead Resting Electrocardiogram Database (PF12RED) was created, containing 163 ECGs in XML format.
• Developed 'ECG Visualizer' software and applied three ML methods to diagnose sinus bradycardia.

Main Results:

• The PF12RED database was successfully established.
• 'ECG Visualizer' software was created, enabling signal processing and analysis.
• Machine learning models demonstrated utility in automating the diagnosis of sinus bradycardia.

Conclusions:

• Artificial intelligence and ML can accurately detect simple arrhythmias in athletes.
• The study provides a valuable ECG dataset and a free software application for athlete cardiac health.
• This work enhances the potential for automated diagnosis of cardiac conditions in professional football players.