Forecasting knee arthroplasty surgery with deep learning : an integrated approach using routine clinical data and
Omar Musbahi1, Thomas A G Hall2, Alexis Alibert2
1MSk Lab, Department of Surgery and Cancer, Imperial College London, London, UK.
Bone & Joint Open
|July 13, 2026
Summary
A new deep learning model integrating clinical and radiological data accurately predicts knee arthroplasty needs and timing. This tool can streamline patient care and surgical planning for knee osteoarthritis.
Area of Science:
- Orthopedics
- Artificial Intelligence
- Medical Imaging
Background:
- Knee osteoarthritis is a leading cause of disability, often requiring surgical intervention like knee arthroplasty.
- Predicting the need and timing for knee arthroplasty is crucial for patient management and healthcare resource allocation.
- Current prediction methods often lack accuracy or fail to integrate diverse patient data effectively.
Purpose of the Study:
- To develop and validate a practical deep learning model for predicting knee arthroplasty (total or partial) in patients with or at risk of knee osteoarthritis.
- To predict the time to surgery for knee arthroplasty using integrated clinical and radiological data.
- To assess the model's performance and generalizability using independent datasets.
Main Methods:
- Utilized data from the Multicentre Osteoarthritis Study (MOST) for training and testing, and the Osteoarthritis Initiative (OAI) for external validation.
- Developed a deep learning model based on DenseNet-201, combining outputs from radiological analysis with clinical data.
- Evaluated model performance using area under the receiver operating characteristic curve (AUROC) and precision-recall curve (AUPRC).
Main Results:
- The integrated model achieved an AUROC of 0.85 and AUPRC of 0.62, outperforming models using single data sources.
- External validation with the OAI dataset confirmed generalizability with an AUROC of 0.79.
- The model demonstrated higher predictive accuracy for surgical interventions within 40 months (AUROC 0.83, AUPRC 0.26).
Conclusions:
- Deep learning models integrating clinical and radiological data show significant potential for predicting knee arthroplasty needs.
- The developed model's robust performance and generalizability can enhance clinical pathways and surgical demand forecasting.
- This facilitates better resource planning and ensures timely surgical access for patients requiring knee arthroplasty.
