Interpretable machine learning model integrating MRI-derived paraspinal muscle parameters for predicting new
Chengming Wang1, Mingcong Gao2,3, Junhao Ye1
1Medical Imaging Center, The First People's Hospital of Foshan (The Affiliated Foshan Hospital of Southern University of Science and Technology), School of Medicine, Southern University of Science and Technology, Guangdong, China.
European Radiology
|May 15, 2026
Summary
Machine learning models using MRI-derived muscle data can predict new vertebral compression fractures after augmentation. This aids in personalized patient management and preventive strategies.
Area of Science:
- Orthopedics
- Radiology
- Data Science
Background:
- New vertebral compression fractures (NVCF) post-vertebral augmentation lead to significant morbidity.
- Current risk stratification methods for NVCF are insufficient.
- The prognostic role of MRI-derived paraspinal muscle characteristics is not well-established.
Purpose of the Study:
- To develop and validate interpretable machine learning (ML) models for predicting NVCF after vertebral augmentation.
- To incorporate MRI-derived paraspinal muscle parameters into predictive models.
- To assess the clinical utility of these models for personalized patient management.
Main Methods:
- A multicenter retrospective study analyzed data from 359 patients undergoing vertebral augmentation.
- Machine learning models were trained using clinical, radiographic, and MRI-derived paraspinal muscle parameters (paraspinal muscle fat infiltration [PMFI] and psoas muscle index [PMI]).
- Model performance was evaluated using AUC, calibration, and decision curve analyses, with SHAP values used for interpretability.
Main Results:
- An interpretable random forest model demonstrated high predictive accuracy for NVCF.
- Key predictors included age, bone mineral density (BMD), PMFI, PMI, kyphotic angle correction rate (KACR), and Genant semiquantitative grade (GSQ).
- The model achieved AUCs of 0.963 (training), 0.913 (test), and 0.837 (external validation).
Conclusions:
- An interpretable ML model integrating MRI-based muscle metrics and conventional factors effectively predicts NVCF post-vertebral augmentation.
- This approach enhances risk stratification and supports personalized treatment strategies.
- The findings suggest potential for improved clinical decision-making and preventive interventions.