Risk prediction of sepsis-associated acute kidney injury: development, validation of a machine learning model with multicenter data.
Source check: Source checked. Not independent scientific confirmation.
Finding: Using MIMIC-IV for development and two external ICU databases, XGBoost predicted sepsis-associated acute kidney injury with AUC 0.72 internally and 0.82/0.84 externally; differences across datasets and the retrospective design limit clinical interpretation.
Technology: Six machine-learning algorithms were optimized with grid search and five-fold cross-validation; XGBoost was selected and interpreted with SHAP.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Six machine-learning algorithms were optimized with grid search and five-fold cross-validation; XGBoost was selected and interpreted with SHAP.
- Test and result
- The retrospective model used MIMIC-IV for development and eICU-CRD and BB-ICU for external assessment. XGBoost AUC was 0.72 internally and 0.82 and 0.84 externally; calibration and decision-curve analyses were also reported.
- Limitation / still unknown
- Performance and AKI incidence differed across data sources. Retrospective prediction does not establish causal benefit, prospective performance or clinical utility of the web calculator.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://link.springer.com/article/10.1186/s12911-026-03620-z
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.