Cross-center prediction of ICU-acquired weakness from early admission data
Source check: Source checked. Not independent scientific confirmation.
Finding: In three Shanghai medical centers, five models predicted ICU-acquired weakness among 535 critically ill patients using variables available within 48 hours of admission. External AUROC was 0.853 for SVM and 0.841 for logistic regression; tree-based models showed overfitting. Model-guided care and improved patient outcomes were not tested.
Technology: Logistic regression, random forest, SVM, CART and XGBoost trained on early ICU clinical variables
Publication: Peer-reviewed paper · Test context: prospective or independent check. Catalog categories, not an evidence grade.
- AI contribution
- Five predictive algorithms were trained with 10-fold cross-validated grid search on data available within 48 hours of admission; a six-predictor logistic model was also converted to a nomogram.
- Test and result
- Five machine-learning algorithms and logistic regression were compared in 395 internal-cohort patients and externally validated in 140 patients from two other centers. External AUROC was 0.853 for SVM and 0.841 for logistic regression; tree-based methods showed substantial performance attenuation consistent with overfitting.
- Limitation / still unknown
- The study did not test whether model-guided care reduces ICU-acquired weakness or improves recovery; prospective interventional evaluation is still needed.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC13744720/ · https://pubmed.ncbi.nlm.nih.gov/43001150/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
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 3/5 · Validation 4/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.