Development and Validation of a Nomogram Predicting Postoperative Recurrent Lumbar Disc Herniation Based on Activity...

Area of Science:

• Neurosurgery
• Orthopedics
• Medical Informatics

Background:

• Postoperative recurrent lumbar disc herniation (PRLDH) poses a significant challenge after percutaneous endoscopic transforaminal discectomy (PETD).
• Predictive models are needed to identify patients at high risk for PRLDH to enable personalized management strategies.

Purpose of the Study:

• To develop and validate an individualized predictive model for PRLDH following PETD.
• The model incorporates postoperative activity factors to enhance prediction accuracy.

Main Methods:

• Retrospective analysis of 612 patients undergoing PETD.
• Lasso regression and random forest models were employed for variable selection and prediction.
• Model performance was evaluated using AUC, calibration, DCA, and AIC, with internal validation via Bootstrap and cross-validation.

Main Results:

• 66 out of 612 patients (10.78%) experienced PRLDH.
• The Lasso regression model (Model 1) demonstrated superior predictive performance (AUC: 0.813 training, 0.868 validation) compared to the random forest model.
• Key predictors identified include BMI, smoking, activity level, time to ambulation, diabetes, Modic changes, and Pfirrmann grade.

Conclusions:

• The developed Nomogram predictive model, based on postoperative activity factors, shows excellent capability in assessing PRLDH risk after PETD.
• This tool can aid clinicians in personalized risk stratification for patients undergoing PETD.