An interpretable machine-learning model for predicting in-hospital mortality in patients with sepsis-associated acute kidney injury - PubMed
5 hours ago
- #acute kidney injury
- #sepsis
- #machine learning
- Study focuses on predicting in-hospital mortality in sepsis-associated acute kidney injury (S-AKI) patients using machine learning.
- Retrospective analysis used MIMIC-IV 3.0 database and a prospective cohort from Ningxia Medical University.
- Five machine learning models were trained, with XGBoost showing superior performance (AUC 0.8799).
- Key predictors included SAPS II score, AKI stage, oxygenation index, and biochemical markers like serum sodium and blood urea nitrogen.
- SHAP analysis enhanced model interpretability, identifying critical clinical variables influencing mortality risk.
- External validation confirmed the robustness and applicability of the XGBoost model for S-AKI prognosis.