Machine learning models for predicting patient satisfaction after adult spinal deformity surgery

Area of Science:

• Spine Surgery
• Orthopedics
• Medical Machine Learning

Background:

• Patient satisfaction is a crucial metric for evaluating adult spinal deformity (ASD) surgical outcomes.
• Predictive factors for satisfaction in ASD patients are not well-established.

Purpose of the Study:

• To develop and validate predictive models for patient satisfaction 24 months post-ASD surgery.
• To identify key indicators influencing patient satisfaction after ASD surgery.

Main Methods:

• Utilized machine learning algorithms (LASSO, RFE, Boruta) on a dataset of 213 ASD patients.
• Developed and validated a logistic regression model using 128 training and 85 testing patients.
• Employed SHAP analysis to determine feature importance for satisfaction prediction.

Main Results:

• Identified 9 key indicators of postoperative satisfaction.
• The predictive model achieved an AUC of 0.846 and accuracy of 0.812 in the test set.
• Significant predictors included improved WOMAC function, absence of frailty, and radiographic improvements.

Conclusions:

• Developed models offer personalized prognostic information for ASD patients.
• Surgeons can utilize these modifiable indicators to guide postoperative rehabilitation and improve patient satisfaction.