25/05/2026
Excited to share our latest research: "AI Meets Offloading: Explainable Machine Learning for Footwear Prescription in Hallux-Affected Plantar Foot Ulceration."
Plantar foot ulceration (PFU) in patients with hallux deformities—such as hallux rigidus, hallux limitus, and hammer toes—carries serious risks, including re-ulceration, infection, and amputation. Despite this, footwear and orthotic prescriptions currently rely heavily on individual clinician experience, with minimal standardisation. Our study aimed to change that.
Using a curated retrospective dataset from a specialised foot clinic, we developed and evaluated explainable machine learning models to support data-driven, personalised offloading footwear prescriptions. Key findings include:
- Ensemble methods (Gradient Boosting, XGBoost, CatBoost) demonstrated consistently strong predictive performance across all prescription targets.
- Robust accuracy was achieved without synthetic data augmentation.
- Transparent, interpretable models clearly link biomechanical and functional risk factors to specific footwear design decisions, keeping clinicians informed and in control.
This work bridges the gap between AI capability and clinical trust. When clinicians can understand why a model recommends a particular prescription, adoption becomes practical rather than theoretical. This represents a significant step toward safer, more consistent care for patients at high risk of foot complications.
Read the full paper: https://www.sciencedirect.com/science/article/pii/S2949953426000299
Foot Balance Technology Orthogenix Charles Sturt University