Related Experiment Video
Updated: Sep 15, 2025
07:43
In Vivo Quantification of Hip Arthrokinematics during Dynamic Weight-bearing Activities using Dual Fluoroscopy
Published on: July 2, 2021
3.2K
Development and Internal Validation of Machine Learning Algorithms for Predicting Subsequent Contralateral Slipped
David P VanEenenaam1, Carter Hall1, Daniel A Maranho2
1Children's Hospital of Philadelphia, Orthopedic Center, Philadelphia, PA, USA.
Summary
Machine learning models can predict contralateral slipped capital femoral epiphysis (SCFE) risk. The epiphyseal cupping ratio (ECR) is a key radiographic predictor, aiding decisions on prophylactic pinning.
Area of Science:
- Orthopedic surgery
- Pediatric orthopedics
- Medical imaging analysis
Background:
- Controversy exists regarding prophylactic pinning of the contralateral hip in unilateral slipped capital femoral epiphysis (SCFE).
- Machine learning (ML) offers potential for identifying complex patterns to predict which patients may benefit from prophylactic pinning.
- Accurate prediction models are needed to guide clinical decisions for unilateral SCFE management.
Purpose of the Study:
- To develop and evaluate a machine learning (ML) predictive model for contralateral SCFE.
- To identify key demographic, clinical, and radiographic risk factors for contralateral SCFE.
- To assess the utility of emerging radiographic measures, like the epiphyseal cupping ratio (ECR), in predicting contralateral SCFE.
Main Methods:
- Retrospective study of 604 patients under 18 with unilateral SCFE and at least 18 months follow-up.
- Data collected included demographics, clinical information, and radiographic measures (e.g., lateral central edge angle [LCEA], ECR).
- Machine learning models were trained and validated using area under the curve (AUC) and weighted F1-score, with feature selection based on significant relationships between patient groups.
Main Results:
- Younger age ( <12.9 years), lower LCEA ( <30.2°), and lower ECR ( <0.23) were significant predictors of contralateral SCFE.
- A combination of these factors indicated a maximum 28.7% risk of contralateral SCFE.
- The final ML model, incorporating age, skeletal maturity, sex, LCEA, and ECR, achieved an AUC of 0.66 and a weighted F1-score of 0.77.
Conclusions:
- The epiphyseal cupping ratio (ECR) is a significant radiographic predictor of contralateral SCFE.
- A preliminary ML model demonstrates promising accuracy in predicting contralateral SCFE, warranting further investigation with larger datasets.
- These findings can inform clinical decision-making regarding prophylactic pinning in unilateral SCFE cases.