ICU sedation, delirium & recovery
Machine-learning algorithms identify sTREM1 as a key biomarker for outcome prediction in critically ill
Authors' conclusion
sTREM-1 is a robust biomarker associated with mortality and kidney-related outcomes in critically ill patients. Its predictive performance were comparable to multiparametric machine learning models and superior to severity scores.
Study context
- Population
- Critically ill ICU patients requiring invasive mechanical ventilation or a vasoactive agent for >24 hFROG-ICU
- Participants
- 2,061 patients
- Comparison
- ML models and sTREM-1 alone vs severity scores for outcome prediction
- Primary endpoint
- Mortality
- Follow-up
- Day 90
Methodological detailsmachine learning (Random Forest, LASSO regression), external validation (MARS cohort)
- Machine learning (Random Forest, LASSO regression)
- External validation (MARS cohort)
Primary result
As reported in the abstract
Prognostic performance for day-90 mortality
| Group | AUC |
|---|---|
| ML models | 0.74 |
| sTREM-1 | 0.72 |
| Severity scores | 0.64 |
AUC
ML models0.74
sTREM-10.72
Severity scores0.64
- ML models vs Severity scores: P < 0.001 — ML models outperformed severity scores
- Day-90 mortality
- 30.1%
Secondary outcomes
External validation (MARS cohort, sepsis and septic shock)
- Findings confirmed in the MARS cohort
Major adverse kidney event (MAKE) in ICU
- Similar results for MAKE prediction
Original publication
Machine-learning algorithms identify sTREM1 as a key biomarker for outcome prediction in critically ill. Crit Care. Published 27 May 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →