Early Predictive Accuracy of Machine Learning for Hemorrhagic Transformation in Acute Ischemic Stroke: Systematic...

Area of Science:

• Neurology
• Medical Informatics
• Biostatistics

Background:

• Hemorrhagic transformation (HT) is a common complication of acute ischemic stroke (AIS) associated with poor prognosis.
• Current tools for early HT risk prediction are inadequate.
• Machine learning (ML) is emerging as a potential tool for predicting HT in AIS.

Purpose of the Study:

• To systematically evaluate the predictive performance of ML models for HT risk in AIS.
• To identify factors influencing the accuracy of ML-based HT prediction.
• To provide evidence-based guidance for developing improved HT prediction tools.

Main Methods:

• A comprehensive literature search was conducted across major medical databases (PubMed, Embase, Web of Science, Cochrane) up to March 2025.
• The Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to assess study quality.
• Subgroup analyses were performed based on treatment, diagnostic criteria, and HT type.

Main Results:

• 83 studies comprising 106 ML models and data from 88,197 AIS patients (9,323 with HT) were included.
• Overall pooled performance: c-index 0.832, sensitivity 0.82, specificity 0.78.
• Combined models utilizing clinical features and radiomics demonstrated superior predictive accuracy.

Conclusions:

• Existing HT prediction methods have limitations in predictive value.
• ML models, especially those integrating clinical data and radiomics, offer significant potential for enhancing HT risk prediction in AIS.
• This meta-analysis provides a foundation for developing more effective clinical HT prediction tools.