A Supervised Learning Approach Electrocardiographic Model for Differentiating Outflow Tract Premature Ventricular...

Area of Science:

• Cardiology
• Medical Imaging and Diagnostics
• Machine Learning in Medicine

Background:

• Accurate localization of premature ventricular complexes (PVCs) from the right ventricular outflow tract (RVOT) and left ventricular outflow tract (LVOT) is crucial for successful ablation.
• Current electrocardiogram (ECG) algorithms for PVC localization face limitations due to anatomical variability.
• There is a need for improved noninvasive methods to differentiate RVOT from LVOT PVCs.

Purpose of the Study:

• To develop a supervised learning model using logistic regression and validated ECG parameters for PVC localization.
• To compare the diagnostic performance of the developed model against seven established ECG algorithms.
• To assess the model's effectiveness in differentiating RVOT from LVOT PVCs.

Main Methods:

• A retrospective analysis of 116 patients with idiopathic outflow tract PVCs who underwent successful ablation.
• Four key ECG parameters were selected using backward stepwise logistic regression.
• Performance was evaluated using receiver-operating characteristic (ROC) curve analysis, Youden index, and accuracy metrics, including subgroup analysis for V3 precordial transition.

Main Results:

• The supervised learning model achieved superior diagnostic accuracy (AUC 0.942 overall, 0.878 in V3 transition subgroup), significantly outperforming comparator algorithms (P < .001).
• The model demonstrated high diagnostic performance with a Youden index of 0.66, 82.7% sensitivity, and 84.4% specificity.
• The model proved effective in differentiating RVOT from LVOT PVCs.

Conclusions:

• The developed supervised learning model offers a more accurate and reliable method for differentiating RVOT from LVOT PVCs compared to existing rule-based ECG algorithms.
• This statistically optimized and interpretable model serves as a valuable noninvasive tool to aid in planning catheter ablation procedures for PVCs.
• Further validation in larger, multicenter studies is recommended to confirm these findings.