Development and Validation of a Predictive Model Based on LASSO Regression: Predicting the Risk of Early Recurrence of Atrial Fibrillation after Radiofrequency Catheter Ablation

Background: Although recurrence rates after radiofrequency catheter ablation (RFCA) in patients with atrial fibrillation (AF) remain high, there are a limited number of novel, high-quality mathematical predictive models that can be used to assess early recurrence after RFCA in patients with AF. Purpose: To identify the preoperative serum biomarkers and clinical characteristics associated with post-RFCA early recurrence of AF and develop a novel risk model based on least absolute shrinkage and selection operator (LASSO) regression to select important variables for predicting the risk of early recurrence of AF after RFCA. Methods: This study collected a dataset of 136 atrial fibrillation patients who underwent RFCA for the first time at Peking University Shenzhen Hospital from May 2016 to July 2022. The dataset included clinical characteristics, laboratory results, medication treatments, and other relevant parameters. LASSO regression was performed on 100 cycles of data. Variables present in at least one of the 100 cycles were selected to determine factors associated with the early recurrence of AF. Then, multivariable logistic regression analysis was applied to build a prediction model introducing the predictors selected from the LASSO regression analysis. A nomogram model for early post-RFCA recurrence in AF patients was developed based on visual analysis of the selected variables. Internal validation was conducted using the bootstrap method with 100 resamples. The model’s discriminatory ability was determined by calculating the area under the curve (AUC), and calibration analysis and decision curve analysis (DCA) were performed on the model. Results: In a 3-month follow-up of AF patients (n = 136) who underwent RFCA, there were 47 recurrences of and 89 non-recurrences of AF after RFCA. P, PLR, RDW, LDL, and CRI-II were associated with early recurrence of AF after RFCA in patients with AF (p < 0.05). We developed a predictive model using LASSO regression, incorporating four robust factors (PLR, RDW, LDL, CRI-II). The AUC of this prediction model was 0.7248 (95% CI 0.6342–0.8155), and the AUC of the internal validation using the bootstrap method was 0.8403 (95% CI 0.7684–0.9122). The model demonstrated a strong predictive capability, along with favorable calibration and clinical applicability. The Hosmer–Lemeshow test indicated that there was good consistency between the predicted and observed values. Additionally, DCA highlighted the model’s advantages in terms of its clinical application. Conclusions: We have developed and validated a risk prediction model for the early recurrence of AF after RFCA, demonstrating strong clinical applicability and diagnostic performance. This model plays a crucial role in guiding physicians in preoperative assessment and clinical decision-making. This novel approach also provides physicians with personalized management recommendations.