Development and Statistical Validation of a Machine Learning Model for Predicting Functional Outcomes at Discharge
Takehiro Kaneoka1, Tomohiro Kawazoe1, Kazuhiro Yamazaki1
1Department of Orthopaedic Surgery, Shunan Memorial Hospital, Kudamatsu, Japan.
Progress in Rehabilitation Medicine
|December 11, 2025
Summary
A machine learning model accurately predicts post-surgery motor function for hip fracture patients. Key predictors include rehabilitation status and cognitive scores, aiding in personalized discharge planning.
Area of Science:
- Orthopedic Surgery
- Rehabilitation Medicine
- Artificial Intelligence in Healthcare
Background:
- Hip fractures are a significant health concern in aging populations.
- Predicting functional recovery after hip arthroplasty is crucial for effective patient management.
- Timely interventions can improve outcomes for patients with hip fractures.
Purpose of the Study:
- To develop a reliable machine learning model for predicting discharge motor function (mFIM scores).
- To identify key predictors of motor recovery in patients undergoing bipolar hip arthroplasty (BHA).
- To enhance clinical decision-making for rehabilitation and discharge planning.
Main Methods:
- Retrospective analysis of 201 patients treated with BHA for femoral neck fractures.
- Development and hyperparameter tuning of six machine learning models.
- Evaluation of feature importance using SHapley Additive exPlanations (SHAP) and multiple regression.
Main Results:
- A Light Gradient Boosting Machine model achieved high predictive performance (R² = 0.84).
- The most influential predictors identified were mFIM score at transfer, Hasegawa Dementia Scale-Revised (HDS-R) score, and pre-fracture independence.
- These findings were consistent with multiple linear regression analysis.
Conclusions:
- A reliable and interpretable machine learning model was developed to predict discharge mFIM scores post-BHA.
- The model's key predictors offer insights into factors influencing motor recovery.
- This tool can assist clinicians in optimizing rehabilitation goals and discharge plans.
