Construction of a machine learning-based risk prediction model for acute stroke in maintenance hemodialysis patients
Source check: Source checked. Not independent scientific confirmation.
Finding: A model developed from 1,100 maintenance-hemodialysis patients predicted acute stroke in a held-out 330-patient test set. XGBoost achieved AUC 0.847 (95% CI 0.791–0.902), sensitivity 0.73 and specificity 0.83. The study is single-center and retrospective; external validation and clinical benefit remain unestablished.
Technology: Seven machine-learning algorithms; XGBoost performed best
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Seven machine-learning algorithms were trained on clinical variables to estimate acute-stroke risk in maintenance-hemodialysis patients; XGBoost was the best-performing model.
- Test and result
- A single-center retrospective dataset of 1,100 patients was split into 770 training and 330 test cases. XGBoost achieved test AUC 0.847 (95% CI 0.791–0.902), sensitivity 0.73 and specificity 0.83.
- Limitation / still unknown
- The study reports no external validation or clinical-impact evaluation; performance beyond this held-out single-center cohort is unknown.
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/PMC13769277/ · https://pubmed.ncbi.nlm.nih.gov/43034765/
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 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.