Objective <p>Adjacent segment degeneration (ASDeg) is a common complication following anterior cervical discectomy and fusion (ACDF). While radiomics has been widely applied in the spinal field, it has not yet been utilized for ASDeg screening in patients after ACDF. This study aimed to develop and validate a predictive model that integrates radiomic features and clinical risk factors using machine learning algorithms to achieve accurate screening for ASDeg following ACDF.</p> Methods <p>A retrospective study was conducted on patients with cervical spondylosis who underwent ACDF at our hospital between January 1, 2019, and December 31, 2023. Based on follow-up data, 90 ASDeg patients and 104 non-ASDeg patients were enrolled and randomly divided into a training cohort and a test cohort at a ratio of 7:3. Radiomic features of the intervertebral discs were extracted from preoperative MRI T2-weighted images, and optimal features were selected using univariate analysis and LASSO regression. By integrating radiomic and clinical features, predictive models were constructed using machine learning algorithms (DT, LR, SVM and XGBoost). The predictive performance of the models was evaluated using the receiver operating characteristic (ROC) curve and calibration curve.</p> Results <p>Clinical, radiomics, and deep learning radiomics (DLR) models were constructed based on 2 clinical features, 14 radiomic features, and 8 DLR features, respectively. In the combined model integrating these features, the XGBoost algorithm demonstrated superior predictive performance compared to other models, achieving AUCs of 0.99 in the training cohort and 0.98 in the test cohort, while its calibration curve showed good agreement between predicted and actual probabilities.</p> Conclusion <p>Radiomics and machine learning modeling based on preoperative MRI T2WI images can effectively achieve accurate screening of high-risk populations for ASDeg after ACDF. This finding requires further confirmation through multicenter external validation.</p>