A Novel Lumbar Disc Herniation Classification Based on Endplate Destruction.

Objective: The purpose of this study was to propose a new classification system based on endplate destruction for lumbar disc herniation to guide the selection of clinical treatment strategies.

Methods: Included in this study were 300 lumbar disc herniation patients who received treatment in our center between January 2016 and March 2023. Basic demographic information, treatment plans, and the Oswestry Disability Index scores and visual analog scale scores were collected. A total of 250 patients were included in the training set. These data were used to evaluate the clinical efficacy of treatment protocols for each type of the new classification system. We performed variable selection using machine learning techniques and constructed a nomogram based on the selected important variables. Finally, a validation set comprising data from 50 patients was used to evaluate the model's performance, with the area under the receiver operating characteristic curve serving as the evaluation metric.

Results: It was found that initial symptoms were the most severe in type A patients, intermediate in type B, and the mildest in type C. Type C patients predominantly chose conservative treatment. Conversely, a higher proportion of type A patients chose surgical treatment. For type B patients, conservative treatment should be the primary approach. Machine learning analysis revealed that age and classification were key factors influencing surgical decision-making in patients. The predictive performance of the nomogram, validated by an area under the curve of 0.74, demonstrated high accuracy.

Conclusions: The novel classification system based on endplate destruction has demonstrated excellent reliability and can provide significant reference value for guiding the formulation of treatment plans.