TruaceTracing the truth around AIMonday, August 24, 2026
Health·The Trace·Dual reading·Published 2026-08-24

TyG-FI-based mortality risk characterization and metabolic-frailty phenotyping in critically ill patients with acute kidney injury

Source article: Triglyceride-glucose frailty index, metabolic-frailty phenotypes, and mortality in critically ill patients with acute kidney injury: Derivation, interpretation, and external validation

Abstract: Background Prognosis remains heterogeneous among critically ill patients with acute kidney injury (AKI). We evaluated the triglyceride-glucose frailty index (TyG-FI), a composite of metabolic burden and laboratory-based frailty, for mortality risk characterization, phenotype identification, prediction, and external validation. Methods We included 2230 adults with KDIGO-defined AKI from MIMIC-IV. Associations between TyG-FI and ICU, in-hospital, 28-day, 90-day, and 365-day mortality were assessed using multivaria…

TRV-2026-0862Peer-reviewedPermanent record — cite & verify
Trace impact reading

Contested: both sides are scored from claims and sources, not community votes.

P 72The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.G 68The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.
Triglyceride-glucose frailty index, metabolic-frailty phenotypes, and mortality in critically ill patients with acute kidney injury: Derivation, interpretation, and external validation

"Bill on dialysis" by fireflythegreat is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0/.

The quick read

Researchers derived and tested the triglyceride-glucose frailty index in 2230 MIMIC-IV adults with KDIGO-defined AKI, examining associations with ICU, in-hospital, 28-day, 90-day and 365-day mortality, identifying two consensus phenotypes, and evaluating 12 prediction algorithms with SHAP and LIME interpretation and external validation in 1831 eICU patients.

The findings matter because ICU-AKI prognosis is heterogeneous and multidimensional risk tools could inform triage and family discussions, but the observed drop in AUROC on external validation and the limited gain over FI-Lab alone leave uncertainty about clinical utility and whether the composite adds value beyond existing frailty measures.

Main points
  • Study included 2230 adults with KDIGO-defined AKI from MIMIC-IV and 1831 eICU patients for external validation.
  • Consensus clustering identified K=2 metabolic-frailty phenotypes with 28-day mortality rates of 14.4% and 31.5% in MIMIC-IV.
  • Twelve prediction algorithms were evaluated using a 7:3 development-test split, with SHAP and LIME for interpretation and double machine learning as exploratory robustness analysis.
Gain

In 2230 MIMIC-IV adults with AKI, higher TyG-FI was independently associated with ICU and 28-day mortality and enabled identification of lower-burden versus high frailty-organ dysfunction phenotypes that reproduced in 1831 eICU patients with high agreement.

Problem

The TyG-FI showed limited incremental predictive advantage over FI-Lab alone and reduced discrimination on external validation, with AUROCs dropping to 0.694 and 0.667 in eICU compared to 0.764 and 0.769 internally.

The rundown

The analysis used multivariable logistic and Cox regression, restricted cubic splines, subgroup and sensitivity analyses to test TyG-FI associations across ICU, in-hospital, 28-day, 90-day, and 365-day mortality endpoints.

Phenotype reproducibility was assessed by independent clustering in eICU and by transport of locked MIMIC-IV centroids, yielding 95.4% agreement between independently derived and transported phenotype assignments.

Double-machine-learning estimates remained positive across sensitivity analyses, supporting robustness of the association between TyG-FI and mortality.

What this doesn’t fix

Incremental predictive value of TyG-FI over frailty alone was limited, and external validation showed lower discrimination than internal testing.

Sources

Reader signal

How should this claim be treated?

The debate