What does the research say about GERI-NutriRisk-ICU?

GERI-NutriRisk-ICU

There is 1 paper on GERI-NutriRisk-ICU in aging research: none in people yet. The newest story, Nov 12, 2026: “Model predicted hospital death risk in older intensive care patients”.

Biomarker

1 paper studies it2 findings

1 new paper in 30 daysSee all 1 papers on GERI-NutriRisk-ICU

“the domain-updated model achieved AUROC 0.872, Brier score 0.073, and expected calibration error 0.0037.”

GERI-NutriRisk-ICU: an APACHE-contextualized, hospital-robust, and nutrition-aware domain-updated framework for mortality risk stratification in older ICU patientsFrontiers in nutrition2026doi:10.3389/fnut.2026.1980118

In silico1

Effects

Mortality prediction performance

upin humans1

1 study

GERI-NutriRisk-ICU is linked with better mortality prediction performance in humans (Compared with recalibrated APACHE-IVa in the same subset) (AUROC difference 0.021, 95% CI 0.016–0.026; p < 0.001).
“GERI-NutriRisk-ICU significantly outperformed recalibrated APACHE-IVa (AUROC 0.871 vs. 0.850; paired difference 0.021, 95% CI 0.016–0.026, p < 0.001;”
HumansCompared with recalibrated APACHE-IVa in the same subsetn = 39,920
Model predicted hospital death risk in older intensive care patients
Frontiers in nutritionNov 12, 2026

Associations

In-hospital mortality

upin humans1

1 study

GERI-NutriRisk-ICU predicts in-hospital mortality in humans (Primary held-out eICU partition across 206 hospitals) (AUROC 0.872; Brier score 0.073; expected calibration error 0.0037).
“the domain-updated model achieved AUROC 0.872, Brier score 0.073, and expected calibration error 0.0037.”
HumansPrimary held-out eICU partition across 206 hospitalsn = 57,488
Model predicted hospital death risk in older intensive care patients
Frontiers in nutritionNov 12, 2026

Latest

1 paper in 12 months

Frontiers in NutritionNov 12

Model predicted hospital death risk in older intensive care patients

The updated model scored 0.872 at distinguishing hospital deaths from survivors among 57,488 older intensive care patients across 206 hospitals.The updated model scored 0.872 at distinguishing hospital deaths from survivors among 57,488 older intensive care patients across 206 hospitals.

Accepted, not yet published: cohort · 57,488 peopleBiomarkers

Paper