Abstract:Objective: To establish and validate a clinical predictive model for the risk of recurrent lumbar disk herniation (rLDH) after percutaneous endoscopic lumbar discectomy (PELD). Methods: Patients with lumbar disc herniation(LDH) who received PELD were retrospectively collected from January 2017 to January 2020 in Minimally Invasive Spine Surgery Center, Affiliated Hospital of Jining Medical University. Basic clinical features and imaging data were collected, eligible patients were classified as the training set and the validation set at a 6:4 ratio,respectively,and each group was further divided into recurrent group and non-recurrent group. On the basis of univariate and multivariate Logistic regression, the independent risk factors were screen out and the clinical prediction model was constructed. The area under curve(AUC) of the receiver operating characteristic(ROC),calibration curve,decision curve analysis (DCA),were used to evaluate the discrimination,calibration,and clinical impact of the model respectively. Results: A total of 286 patients were enrolled in this study. Logistic regression analysis showed that Modic changes, the range of motion, course of disease and BMI index were the independent risk factors for postoperative recurrence, and a nomogram was successfully constructed. It was verified that the area under ROC curve of the training set was 0.748 (95%CI: 0.490-0.897), and the area under ROC curve of the verification set was 0.767 (95%CI: 0.778-0.667). Calibration curve and Hosmer-Lemeshow test show that the trend of model prediction curve is close to the actual curve when the prediction probability is 5%~70%. When the risk threshold of the training set is between 5% and 80% and the risk threshold of the verification set is between 5% and 55%, the model can produce a large net benefit. Conclusions: The prediction model based on Modic changes of lumbar intervertebral disc, the range of motion, course of disease and BMI index showed the great predictive efficacy, which can help clinicians to assess the risk of rLDH.