Deep learning unmasks the ECG signature of Brugada syndrome

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence

Background:

• Inherited arrhythmic cardiomyopathies, like Brugada Syndrome (BrS), cause sudden cardiac death (SCD) in 10% of cases, often without warning.
• Standard electrocardiograms (ECGs) may not reveal BrS due to normal physiological variations, complicating diagnosis.
• Current diagnostic methods, such as sodium channel blockers, carry significant proarrhythmic risks.

Purpose of the Study:

• To develop and validate a machine-learning algorithm for non-intrusive detection of Brugada Syndrome from ECG data.
• To improve early identification of individuals at risk for life-threatening arrhythmias and sudden cardiac death.

Main Methods:

• A machine-learning algorithm was developed to extract, align, and classify ECG waveforms.
• The algorithm was trained and validated to identify features indicative of BrS.
• The protocol was specifically designed to function without the administration of sodium channel blockers.

Main Results:

• The machine-learning algorithm achieved 88.4% accuracy in identifying Brugada Syndrome.
• The algorithm demonstrated a high discriminative power with an Area Under the Curve (AUC) of 0.934 in validation.
• The developed protocol successfully identified BrS without the need for pharmacologic challenge.

Conclusions:

• This machine-learning approach offers a safe and effective method for diagnosing Brugada Syndrome.
• The algorithm can assist clinicians in identifying patients with this potentially fatal inherited heart condition.
• This non-intrusive diagnostic tool has the potential to reduce the incidence of sudden cardiac death from BrS.