An ensemble machine learning-based risk stratification tool for 30-day mortality prediction in critically ill cardiovascular patients
Authors' conclusion
The ensemble ML model outperformed conventional prognostic tools in predicting 30-day mortality, with SHR augmenting traditional tools but not the ensemble ML model. This approach offers a reliable, interpretable framework for risk stratification in high-risk cardiovascular patients.
Study context
- Population
- ICU patients with CVD combined with DM (2008–2022)
- Participants
- 1,595 patients
- Exposure
- Ensemble machine learning (ML) model vs individual ML models and conventional severity scores
- Outcome
- 30-day mortality prediction (AUC)
- Follow-up
- 30 days
Methodological detailsprognostic model development and validation, ensemble machine learning, external validation
- Prognostic model development and validation
- Ensemble machine learning
- External validation
Main finding
As reported in the abstract
30-day mortality prediction, AUC
| Group | Value | AUC |
|---|---|---|
| Ensemble model | 0.912 (95% CI 0.888–0.936) | 0.912 |
| XGBoost | 0.903 | 0.903 |
| APS III/SOFA/SAPS II | ≤ 0.742 | ≤ 0.742 |
Value
Ensemble model0.912 (95% CI 0.888–0.936)
XGBoost0.903
APS III/SOFA/SAPS II≤ 0.742
AUC
Ensemble model0.912
XGBoost0.903
APS III/SOFA/SAPS II≤ 0.742
- Ensemble model vs XGBoost and APS III/SOFA/SAPS II: all P < 0.001 — ensemble outperformed
Observational data: associations, not effects.
Further findings
30-day mortality rate
- 30-day mortality, entire cohort
- 10.8% (n = 173)
Stress hyperglycemia ratio (SHR) added to SOFA, AUC
- SOFA AUC increased with SHR
- 0.741 to 0.757, P = 0.010
SHR added to ensemble model
- Limited incremental benefit, ΔAUC
- - 0.032, P = 0.094
External validation, AUC
- Ensemble model AUC in independent cohort (n = 307)
- 0.891 (95% CI: 0.864-0.917)
Top predictors
- Anti-hypertensives, aspirin, blood urea nitrogen (BUN), white blood cell (WBC), age, red blood cell (RBC)
Original publication
An ensemble machine learning-based risk stratification tool for 30-day mortality prediction in critically ill cardiovascular patients. Cardiovasc Diabetol. Published 30 September 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →