An interpretable machine learning model for predicting postoperative hypotension in type 2 diabetes mellitus undergoing non‑cardiac surgery
Authors' conclusion
An interpretable ML model based on routinely collected clinical data accurately predicts POH risk in T2DM patients after non-cardiac surgery. The model combines strong discriminative performance with clinical explainability, suggesting its potential as a practical tool for preoperative risk stratification and personalized postoperative monitoring in T2DM patients within similar clinical settings.
Study context
- Population
- T2D patients undergoing non-cardiac surgery; Postoperative hypotension (POH) defined as systolic BP < 90 mmHg in the post-anaesthesia care unit (PACU) (Perioperative Quality Initiative (POQI) consensus)34,012 retrospective, 10,528 prospective
- Outcome
- POH during PACU stay (systolic BP < 90 mmHg after leaving the operating theatre and before transfer to the general ward)
Methodological detailssingle-centre cohort, retrospective and prospective cohorts, machine learning, external validation, SHAP interpretability
- Single-centre cohort
- Retrospective and prospective cohorts
- Machine learning
- External validation
- SHAP interpretability
Performance
As reported in the abstract
Random Forest model discrimination (AUC)
- AUC, training
- 0.843 (95% CI 0.837-0.849)
- AUC, internal validation
- 0.854 (95% CI 0.848-0.860)
- AUC, prospective validation
- 0.847 (95% CI 0.840-0.854)
- AUC, external validation, independent hospital cohort n = 2156
- 0.822 (95% CI 0.805-0.839)
- Sensitivity
- 0.932
Further findings
Top predictors (SHAP)
- Intraoperative blood loss, age, HF, obstructive sleep apnoea and BMI were the top predictors
Original publication
An interpretable machine learning model for predicting postoperative hypotension in type 2 diabetes mellitus undergoing non‑cardiac surgery. Cardiovasc Diabetol. Published 22 June 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →