Objective:

To develop and externally validate explainable machine-learning models for perioperative surgical site infection (SSI) risk enrichment after operative treatment of closed pilon fractures, with a focus on clinical interpretability.

Approach:

• Study Design: Retrospective multicenter study including 1,876 patients treated at an internal center between January 2020 and December 2024 and 359 from an external center treated between August 2024 and September 2025.

Key Findings:

• Internal cohort had 74 SSI events (3.9%), and external cohort had 11 SSI events (3.1%).
• RF model showed highest discrimination with ROC-AUCs of 0.899 (95% CI, 0.831–0.953) (internal) and 0.902 (95% CI, 0.707–0.991) (external).
• RF specificity was high at 0.987 (95% CI, 0.976–0.996) (internal) and 0.974 (95% CI, 0.957–0.989) (external).
• Preoperative-only RF model achieved ROC-AUCs of 0.884 (95% CI, 0.800–0.968) (internal) and 0.905 (95% CI, 0.800–0.968) (external).

Interpretation:

The machine-learning framework demonstrated favorable discrimination for perioperative SSI risk enrichment, with RF as the primary high-specificity model, but further validation is necessary.

Limitations:

• Limited RF sensitivity suggests it should not be viewed as a rule-out system, indicating a need for cautious interpretation.
• Calibration uncertainty indicates the need for further validation before clinical use.

Conclusion:

The study presents a machine-learning approach for SSI risk assessment in closed pilon fracture surgeries, but further recalibration and prospective validation are necessary.