Interpretable machine learning model for predicting recurrence in patients with diabetic foot ulcers
Authors' conclusion
The XGBoost model demonstrated strong predictive accuracy and clinical relevance in assessing DFU recurrence risk. However, further multicenter validation with a larger sample size is needed to improve its generalizability and clinical applicability.
Study context
- Population
- Patients with diabetic foot ulcer (DFU)training/test 4:1
- Participants
- 494 patients
- Comparison
- Seven machine learning (ML) algorithms for predicting 3-year DFU recurrence
- Outcome
- Predictive performance (AUROC) for DFU recurrence
- Follow-up
- 3 years
Methodological detailsmachine learning, training and test sets at 4:1 ratio, fivefold cross-validation, Platt scaling calibration, SHAP interpretability analysis
- Machine learning
- Training and test sets at 4:1 ratio
- Fivefold cross-validation
- Platt scaling calibration
- SHAP interpretability analysis
Performance
As reported in the abstract
AUROC for 3-year DFU recurrence
| Test or model | AUROC | Brier score after Platt scaling |
|---|---|---|
| XGBoost | 0.924 (95% CI 0.867 to 0.967) | 0.096 |
XGBoost
AUROC0.924 (95% CI 0.867 to 0.967)
Brier score after Platt scaling0.096
Further findings
SHAP analysis
- Key risk factors identified aligned with existing literature and clinical expertise
Original publication
Interpretable machine learning model for predicting recurrence in patients with diabetic foot ulcers. BMJ Open Diabetes Res Care. Published 12 November 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →