Machine Learning Models Enhance Prediction of Arrhythmogenic Right Ventricular Cardiomyopathy.

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Arrhythmogenic Right Ventricular Cardiomyopathy (ARVC) is a significant cause of sudden cardiac death globally.
• Current ARVC diagnosis is complex, costly, and time-consuming, delaying clinical decisions.
• Machine learning (ML) presents a potential solution for rapid and scalable ARVC detection.

Purpose of the Study:

• To benchmark eight ML algorithms for ARVC detection.
• To identify the most effective ML model for ARVC diagnosis.
• To evaluate the potential of ML as a clinical decision-support tool.

Main Methods:

• Eight ML algorithms were evaluated for ARVC detection.
• Area-under-the-curve (AUC) and accuracy were the primary performance metrics.
• Rigorous cross-validation was employed to ensure model robustness.

Main Results:

• Gradient Boosted Trees (GBT) demonstrated superior performance compared to other ML models.
• The GBT classifier achieved an accuracy of 94.34% in ARVC detection.
• ML models offer rapid and scalable predictions for ARVC.

Conclusions:

• Gradient Boosted Trees show significant promise as an effective decision-support tool for ARVC diagnosis.
• Implementing ML, particularly GBT, can streamline the ARVC evaluation process.
• This approach has the potential to improve patient outcomes by accelerating diagnosis and treatment.