Deep Learning Model for Predicting Rhythm Outcomes after Radiofrequency Catheter Ablation in Patients with Atrial Fibrillation

Figures

• 

  Figure 1. Flowchart of the experimental procedure. Of the 15 factors, 11 were selected using the extreme gradient boosting (XGBoost) algorithm. We used 4-fold cross-validation for model evaluation and applied the synthetic minority oversampling technique to the training set in each fold. ML, machine learning; MLP, multilayer perceptron; PVI, pulmonary vein isolation.

• 

  Figure 2. Network architecture based on multilayer perceptron (MLP) to predict the late recurrence probability. Dim, dimension; ReLU, rectified linear unit.

• 

  Figure 3. Importance of features obtained from extreme gradient boosting algorithm.AF, atrial fibrillation; LV, left ventricular; LVEF, left ventricular ejection fraction; LA, left atrium; eGFR, estimated glomerular filtration rate; HTN, hypertension; AF type, paroxysmal AF vs. persistent AF; DM, diabetes mellitus; HF, heart failure; CAD, coronary artery disease.

• 

  Figure 4. Confusion matrices of the proposed model for 4-fold cross-validation.

• 

  Figure 5. Receiver operating characteristic (ROC) curves for late recurrence of atrial fibrillation after radiofrequency catheter ablation obtained from the proposed model stratified over 4-fold cross-validation.

• 

  Figure 6. Confusion matrices of conventional prediction models and machine learning approaches. APPLE, acute patient physiologic and laboratory evaluation.

• 

  Figure 7. Receiver operating characteristic (ROC) curves for late recurrence of atrial fibrillation after radiofrequency catheter ablation obtained from conventional prediction models and machine learning approaches. APPLE, acute patient physiologic and laboratory evaluation; XGBoost, extreme gradient boosting; SVM, support vector machine.

• 

  Figure 8. The Shapley additive explanation (SHAP) summary plot of the proposed model. It represents the feature importance of the model output. The color of the dots indicates the attribution value of the feature. For categorical features (CAD {0, 1}; AF type {0: paroxysmal, 1: persistent}; DM {0, 1}; HTN {0, 1}; sex {0: female, 1: male}; stroke {0, 1}; HF {0, 1}), the red and blue dots represent 1 and 0, respectively. AF, atrial fibrillation; LV, left ventricular; LVEF, left ventricular ejection fraction; LA, left atrium; eGFR, estimated glomerular filtration rate; HTN, hypertension; AF type, paroxysmal AF vs persistent AF; DM, diabetes mellitus; HF, heart failure; CAD, coronary artery disease.