Performance of Machine Learning for Tissue Outcome Prediction in Acute Ischemic Stroke: A Systematic Review and Meta-Analysis
Problem
Predicting final infarcted brain tissue in acute ischemic stroke from admission CT or MRI, which is needed to guide treatment decisions.
Why it matters
Accurate infarct prediction directly affects treatment choices and patient outcomes in stroke care.
What the AI did
A systematic review and meta-analysis of 11 studies of machine learning models that predict final infarct tissue from baseline imaging, pooling their dice similarity coefficients.
Results
Pooled DSC 0.50 (95% CI 0.39–0.61) with high heterogeneity (I2 96.5%, p < 0.001)
Impact
Evidence synthesis shows ML infarct prediction performance remains modest and highly heterogeneous across studies.
Record
| Venue | Frontiers in Neurology |
|---|---|
| Published | 2022 |
| Work type | review |
| Authors | Xinrui Wang, Yi‐Ming Fan, Nan Zhang, Jing Li, Yang Duan, Benqiang Yang |
| Citations | 23 (OpenAlex, at import time) |
| References | 61 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3389/fneur.2022.910259 |
| Open access | gold |
| Keywords | machine learning, acute ischemic stroke, infarct prediction, meta-analysis, dice similarity coefficient |
Importance rating
significant — thousands to millions affected
Final infarct prediction guides acute stroke treatment, a clinically significant decision point, but abstract gives no scale of affected population.