Prediction of Clinically Significant Improvement after Lumbar Fusion Surgery Based on Machine Learning

Lumbar fusion surgeries have increased substantially, making patient satisfaction an important indicator of surgical outcomes and quality of care. Conventional statistical approaches have limitations in predicting clinically significant improvement (CSI). This study aimed to develop a machine learning model to predict CSI after lumbar fusion surgery using only preoperative factors to support clinical decision-making.

A total of 359 patients who underwent lumbar fusion surgery between January 2021 and December 2023 were included. Twenty-two preoperative variables, including demographic characteristics and comorbidities, were used for model development. A multi-label classification approach was applied to predict improvements in the 36-item short-form survey (SF-36) mental component summary (MCS) and physical component summary (PCS). CSI was defined as improvement when the average postoperative SF-36 score (PCS or MCS) across available follow-up time points exceeded the preoperative baseline value. Six machine learning algorithms were evaluated using 5-fold cross-validation. Model performance was assessed using seven evaluation metrics, with emphasis on the F1-score and the area under the receiver operating characteristic curve (AUROC). Model interpretability was examined using SHapley Additive exPlanations (SHAP) analysis and waterfall plots.

The Extra Trees model achieved the best performance, with an F1-score of 0.850 and an AUROC of 0.835. SHAP analysis identified hypertension, diabetes mellitus, body mass index, bone mineral density, and revision status as key predictors. Postoperative complications and revision surgery were also analyzed for their associations with comorbidities and outcomes.

The machine learning model accurately predicts CSI after lumbar fusion surgery using preoperative factors and may assist clinicians and patients in preoperative decision-making.