Kissing Spine and Other Imaging Predictors of Postoperative Cement Displacement Following Percutaneous Kyphoplasty: A Machine Learning Approach.

Abstract

To investigate the risk factors associated with postoperative cement displacement following percutaneous kyphoplasty in patients with osteoporotic vertebral compression fractures and to develop predictive models for clinical risk assessment. This retrospective study included 198 patients with osteoporotic vertebral compression fracture who underwent percutaneous kyphoplasty. Imaging and clinical variables were collected. Multiple machine learning models, including logistic regression (LR), L1-and L2-regularized LR, support vector machine (SVM), decision tree, gradient boosting, and random forest, were developed to predict cement displacement. L1-and L2-regularized LR models identified 4 key risk factors: kissing spine (L1: 1.11; L2: 0.91), incomplete anterior cortex (L1: -1.60; L2: -1.62), low vertebral body computed tomography (CT) value (L1: -2.38; L2: -1.71), and large Cobb change (L1: 0.89; L2: 0.87). The SVM model achieved the best performance (accuracy: 0.983, precision: 0.875, recall: 1.000, F1-score: 0.933, specificity: 0.981, area under the curve: 0.997). Other models, including LR, decision tree, gradient boosting, and random forest, also showed high performance but were slightly inferior to SVM. Key predictors of cement displacement were identified, and machine learning models were developed for risk assessment. These findings can assist clinicians in identifying high-risk patients, optimizing treatment strategies, and improving patient outcomes.