Related Experiment Video
Updated: Jun 9, 2026
06:28
Modeling Stroke in Mice - Middle Cerebral Artery Occlusion with the Filament Model
Published on: January 6, 2011
95.3K
Predictive performance of machine learning models in acute ischemic stroke: a systematic review and meta-analysis
Uzma Khanum1, Vasudeva Guddattu2, Shasthara Paneyala3
1Division of Medical Statistics, School of Life Sciences, JSS Academy of Higher Education and Research, Mysuru, India.
Frontiers in Neurology
|March 27, 2026
Summary
Machine learning (ML) models show promise for predicting acute ischemic stroke (AIS) outcomes, achieving a pooled AUC of 0.87. However, heterogeneity and limitations necessitate further research for reliable stroke care management.
Area of Science:
- Neurology
- Medical Informatics
- Biostatistics
Background:
- Acute ischemic stroke (AIS) is a major global health challenge, causing significant mortality and disability.
- Machine learning (ML) models offer advanced analytical capabilities for complex clinical data in AIS prognosis.
Purpose of the Study:
- To systematically review and analyze current ML models for AIS, identifying gaps, methodological quality, and performance.
- To compare the effectiveness of frequently used ML algorithms in AIS prediction.
- To guide future development of ML-based predictive models for stroke care.
Main Methods:
- Systematic review adhering to PRISMA guidelines, registered with PROSPERO (CRD420251033217).
- Searches conducted in PubMed, Scopus, and Web of Science.
- Quality and bias assessment using PROBAST and TRIPOD-AI.
- Meta-analysis of AUC values using a random-effects model with SPSS and R-Studio.
Main Results:
- 14 studies were included, with 12 eligible for meta-analysis.
- Pooled AUC for ML models was 0.87 (95% CI, 0.83-0.91), indicating strong predictive performance.
- Random Forest (AUC=0.85) and SVM (AUC=0.82) outperformed Logistic Regression (AUC=0.75); XGBoost showed AUC=0.82.
- Substantial heterogeneity (I²=99%) was observed, driven by study design, publication year, and algorithm type.
Conclusions:
- ML models demonstrate significant potential for improving prognostic accuracy in AIS.
- Methodological limitations and heterogeneity across studies restrict the generalizability of current findings.
- Further research is needed to refine ML models and enhance their clinical utility in stroke care.
