Background: Atrial fibrillation (AF) is a common cardiac arrhythmia that affects millions of people worldwide. We aim to investigate how to improve prognostic value of recurrence for atrial fibrillation patients after cryoablation by non-linear survival models. Methods: In this study, we retrospectively reviewed data from 1023 patients who underwent cryoablation surgery for AF at Fujian Provincial Hospital (FPH). We generated radiomics signatures (RSI) and a clinical signature (CLI) using a non-linear survival model by repeated 10-fold cross-validation. The comprehensive risk score (TCRS) was obtained by linearly weighting the multivariate Cox proportional risk model. Results: The combination of RSI and CLI indicators had a significantly higher area under the curve (AUC) in the ROC curve of the training set (AUC=0.955) compared to the AUC of a single indicator CLI (AUC=0.862). The TCRS showed better prognostic performance compared to the traditional Lasso-Cox models, with AUC of 0.955 vs 0.664. The accuracy of the model was further confirmed by the C-indices of RSI (C-index: 0.8894; 95%CI: 0.8166-0.9621), CLI (C-index: 0.8431; 95%CI: 0.7466-0.9395), and TCRS (C-index: 0.9072; 95%CI: 0.8281-0.9864) in validation set 2. Conclusions: Under a nonlinear survival model, TCRS which combines RSI and CLI indicators has potential as a promising prognostic tool for post-cryoablation AF patients.