Wound infection is a major determinant of poor prognosis in patients with diabetic foot ulcers (DFUs) and contributes to higher risks of hospitalization and limb complications. Identifying individuals at greater risk of infection may support earlier clinical intervention. A study published in Diabetes, Metabolic Syndrome and Obesity evaluated the performance of multiple machine learning (ML) models designed to predict wound infection in patients with DFUs using routinely collected clinical indicators.

The retrospective analysis included 800 patients with DFUs. A primary cohort of 500 patients was randomly divided into a training dataset (70%, n = 350) and an internal testing dataset (30%, n = 150). An additional independent cohort of 300 patients served as an external validation dataset. Eight ML algorithms were constructed and compared: logistic regression, decision tree, random forest, support vector machine, k-nearest neighbor, naive Bayes, extreme gradient boosting, and light gradient boosting machine. Model performance was assessed using area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, and other metrics in internal cross validation and external validation analyses. SHapley Additive exPlanations (SHAP) were used to examine the relative contribution of predictive variables.

Among the evaluated algorithms, the random forest model demonstrated the strongest predictive performance. The model achieved an AUC of 0.937 (95% CI 0.906 to 0.969) in the training dataset, 0.853 (95% CI 0.804 to 0.901) in the internal testing dataset, and 0.832 (95% CI 0.779 to 0.885) in the external validation dataset. SHAP analysis identified six variables as the most influential predictors of wound infection: age, duration of diabetes, ankle-brachial index, ulcer area, vascular complications, and osteomyelitis.

The analysis indicates that the random forest model demonstrated strong predictive performance and consistent discrimination across development and validation datasets for infection risk in patients with DFUs. Integration of such ML-based tools into clinical workflows may support earlier risk stratification and more individualized management strategies.