STOMAPY: Artificial Intelligence for Risk Stratification of Outcomes Requiring Enterostomal Therapy After Hospital Discharge Following Colorectal Surgery

Introduction: Infectious and wound-healing complications after colorectal surgery often increase the complexity of local care and the need for specialized enterostomal therapy follow-up after hospital discharge. Despite the growing use of predictive models in digestive surgery, a translational gap remains between perioperative prediction and the practical organization of specialized care. Therefore, the aim of this study was to develop and temporally validate a machine-learning-based risk stratification model to estimate the probability of post-discharge outcomes associated with greater demand for enterostomal therapy after colorectal surgery. Methods: This was a retrospective observational study including 7,908 patients who underwent colorectal surgery between 2005 and 2014. The outcome was defined as the occurrence of superficial surgical site infection, delayed wound healing, or abdominal sinus formation. Routinely available preoperative and intraoperative variables were used as predictors. The primary model was based on gradient boosting with isotonic calibration. Temporal validation was performed by separating cohorts according to year of surgery. Performance was assessed using ROC-AUC, PR-AUC, Brier score, calibration, and decision-oriented clinical metrics. Clinical utility was examined through percentile-based risk stratification and Decision Curve Analysis (DCA). Results: The outcome prevalence in the test set was 6.6%. The calibrated model achieved a ROC-AUC of 0.64 and a PR-AUC of 0.11, with a Brier score of 0.061. The Top-10% risk stratum concentrated approximately twice the baseline event rate ({approx}14% vs. 6.6%), with a number needed for intensified follow-up of 7 patients to identify one event. Decision curve analysis showed greater net benefit than strategies of following all or no patients, particularly for threshold probabilities between 3% and 13%. Models based exclusively on preoperative or intraoperative variables performed worse than the combined model. Conclusion: STOMAPY demonstrated the ability to organize patients along a continuous gradient of risk for post-discharge outcomes associated with greater demand for enterostomal therapy. Although discriminatory performance was moderate, the adequate calibration, temporal validation, and net benefit observed across clinically plausible thresholds support its usefulness as a tool for proportional care prioritization rather than as an individual diagnostic test. Prospective studies and external validations are needed to confirm direct clinical impact.