Machine learning-based analysis of risk factors for atrial fibrillation recurrence after Cox-Maze IV procedure in patients with atrial fibrillation and chronic valvular disease: A retrospective cohort study with a control group

Abstract

ObjectivesTo evaluate the efficacy of the Cox-Maze IV procedure (CMP-IV) in combination with valve surgery in patients with both atrial fibrillation (AF) and valvular disease and use machine learning algorithms to identify potential risk factors of AF recurrence.MethodsA total of 1,026 patients with AF and valvular disease from two hospitals were included in the study. 555 patients received the CMP-IV procedure in addition to valve surgery and left atrial appendage ligation (CMP-IV group), while 471 patients only received valve surgery and left atrial appendage ligation (Non-CMP-IV group). Kaplan–Meier analysis was used to calculate the sinus rhythm maintenance rate. 58 variables were selected as variables for each group and 10 machine learning models were developed respectively. The performance of the models was evaluated using five-fold cross-validation and metrics including F1 score, accuracy, precision, and recall. The four best-performing models for each group were selected for further analysis, including feature importance evaluation and SHAP analysis.ResultsThe 5-year sinus rhythm maintenance rate in the CMP-IV group was 82.13% (95% CI: 78.51%, 85.93%), while in the Non-CMP-IV group, it was 13.40% (95% CI: 10.44%, 17.20%). The eXtreme Gradient Boosting (XGBoost), LightGBM, Category Boosting (CatBoost) and Random Fores (RF) models performed the best in the CMP-IV group, with area under the curve (AUC) values of 0.768 (95% CI: 0.742, 0.786), 0.766 (95% CI: 0.744, 0.792), 0.762 (95% CI: 0.723, 0.801), and 0.732 (95% CI: 0.701, 0.763), respectively. In the Non-CMP-IV group, the LightGBM, XGBoost, CatBoost and RF models performed the best, with AUC values of 0.738 (95% CI: 0.699, 0.777), 0.732 (95% CI: 0.694, 0.770), 0.724 (95% CI: 0.668, 0.789), and 0.716 (95% CI: 0.656, 0.774), respectively. Analysis of feature importance and SHAP revealed that duration of AF, preoperative left ventricular ejection fraction, postoperative heart rhythm, preoperative neutrophil-lymphocyte ratio, preoperative left atrial diameter and heart rate were significant factors in AF recurrence.ConclusionCMP-IV is effective in treating AF and multiple machine learning models were successfully developed, and several risk factors were identified for AF recurrence, which may aid clinical decision-making and optimize the individual surgical management of AF.

Comparison of short-term efficacy of two bipolar radiofrequency ablation forceps for rheumatic heart disease concomitant with atrial fibrillation

“…Unfortunately, drug therapy generally has serious side effects, and the stable maintenance of SR after catheter ablation often requires multiple ablations, both of which have not significantly improved patients’ left heart function and quality of life ( 13 ). Currently, the radiofrequency-based Cox-Maze IV procedure has replaced the original “cut-and-sew technique” as the main treatment for surgical ablation of AF, and the corresponding surgical ablation devices have also been developed rapidly ( 14 ). After decades of turnover, in the current domestic market the bipolar radiofrequency ablation forceps ( MZ-RFK ) manufactured by Med-Zenith are more frequently used in clinical application.…”

Novel Entropy-Based Metrics for Long-Term Atrial Fibrillation Recurrence Prediction Following Surgical Ablation: Insights from Preoperative Electrocardiographic Analysis

“…It is worth noting that this classification model, with Acc and AUC values of about 80% and 78%, respectively, exhibited comparable or slightly better performance compared to previous works, even when using a more unbiased and robust validation methodology, such as 100 repetitions in a 5 fold cross-validation approach [47] (see Table 3). Although some aspects related to datasets, like the size, heterogeneity, and follow-up period, might impact the classification results of predictive models, it is still possible to highlight that the proposed ECG-based analysis of the f -wave variability over time can provide useful and complementary information for the previously introduced clinical and anatomical predictors of AF recurrence after Cox-Maze surgery [21][22][23][24]. This idea has also been suggested by Kakuta and coworkers [50], who recently proposed a 10-point risk score model for preoperative predictions of Cryo-Maze success in the long term by integrating clinical and anatomical information with an f -wave characteristic measured from the surface ECG.…”

“…To shed more light on this aspect, in a recent work by Jiang and colleagues [ 24 ], the authors explored multiple machine learning models, combining 58 clinical variables to predict AF recurrence 5 years after the Cox–Maze procedure and valve surgery. The proposed models exhibited AUC values between 73.20% and 76.80%, revealing that the most significant preoperative features were the AF duration, left ventricular ejection fraction, neutrophil–lymphocyte ratio, left atrial diameter, and heart rate.…”

“…Most previous studies dealing with preoperative predictions of Cox–Maze surgery have presented limited outcomes from the point of view of clinical application, emphasizing clinical and anatomical information associated with atrial function [ 21 , 22 , 23 , 24 ]. In particular, Jiang and colleagues [ 24 ] developed multiple machine learning models based on 58 clinical variables, demonstrating that the most significant preoperative features for predicting AF recurrence were the duration of the arrhythmia, left ventricular ejection fraction, left atrial diameter, neutrophil–lymphocyte ratio, and heart rate.…”

“…Most previous studies dealing with preoperative predictions of Cox–Maze surgery have presented limited outcomes from the point of view of clinical application, emphasizing clinical and anatomical information associated with atrial function [ 21 , 22 , 23 , 24 ]. In particular, Jiang and colleagues [ 24 ] developed multiple machine learning models based on 58 clinical variables, demonstrating that the most significant preoperative features for predicting AF recurrence were the duration of the arrhythmia, left ventricular ejection fraction, left atrial diameter, neutrophil–lymphocyte ratio, and heart rate. Other studies have also extolled the high predictive capacity of some of these metrics, e.g., AF duration and preoperative left atrial diameter, yielding accuracy (Acc) values of about 75% [ 22 ], whereas others yielded much lower Acc values of 55.3 and 68.7%, respectively, and highlighted the performance of other metrics such as the heart rate, with an Acc of 75.8% [ 23 ].…”