ABSTRACT
【Objective】 To explore the establishment of individualized prediction model of recurrence after percutaneous endoscopic lumbar discectomy （PELD） in patients with lumbar disc herniation （LDH）. 【Methods】 We selected 124 LDH patients treated with PELD in Department of Orthopedics， The First Affiliated Hospital of Xi’an Jiaotong University， from January 2017 to January 2020 as the research subjects. Their clinical data were retrospectively analyzed， and the independent risk factors affecting PELD recurrence in the LDH patients were screened by univariate analysis and Logistic regression analysis， respectively； the correlation histogram prediction model was established. 【Results】 Age， history of diabetes， course of disease， work intensity and IDDG were the risk factors for the recurrence of PELD in LDH patients （P<0.05）. Based on the risk factors screened out， the prediction model of the histogram was established， and the model was verified. The results showed that the C-index of the modeling set and the validation set was 0.944 （95% CI： 0.902-0.963） and 0.969 （95% CI： 0.911-0.978）， respectively. The correction curves of both groups were well fitted with the standard curves. The areas under the ROC curve （AUC） in the two groups were 0.944 and 0.969， respectively， which proved that the model had good prediction accuracy. 【Conclusion】 LDH patients have many independent risk factors for recurrence after PELD， and the model based on risk factors with good predictive ability can be useful in preoperative evaluation， appropriate patient selection， and decrease of recurrence rate after PELD.