Advanced Machine Learning Models for Predicting Post-Thrombolysis Hemorrhagic Transformation in Acute Ischemic Stroke...

Area of Science:

• Neurology
• Artificial Intelligence
• Medical Informatics

Background:

• Thrombolytic therapy is crucial for acute ischemic stroke (AIS) but carries a risk of hemorrhagic transformation (HT).
• Accurate prediction of HT is vital for optimizing patient management and outcomes in AIS.
• Current prediction methods require enhancement to effectively stratify HT risk.

Purpose of the Study:

• To systematically review and evaluate the performance of machine learning (ML) and deep learning (DL) models in predicting HT in AIS patients treated with thrombolysis.
• To identify key predictors of HT identified by ML/DL models.
• To assess the overall predictive accuracy and clinical utility of these AI-driven approaches.

Main Methods:

• A comprehensive literature search was conducted across major databases (PubMed, Web of Science, Scopus, Embase, Google Scholar) until July 10, 2024.
• Included studies utilized ML/DL algorithms for HT prediction in thrombolyzed AIS patients, adhering to PRISMA guidelines.
• Data extraction and quality assessment were performed using TRIPOD and PROBAST tools.

Main Results:

• Twelve studies involving 18,007 AIS patients were analyzed, demonstrating high predictive performance for ML models (AUC 0.79-0.95).
• XGBoost (AUC up to 0.953) and Artificial Neural Networks (ANN) (AUC up to 0.942) showed superior predictive capabilities.
• Key predictors included age, glucose levels, NIH Stroke Scale (NIHSS) score, blood pressure, and radiomic features.

Conclusions:

• ML techniques, particularly XGBoost and ANN, show significant promise for predicting HT in AIS patients post-thrombolysis.
• These models can enhance risk stratification and inform clinical decision-making for improved AIS management.
• Future research should prioritize prospective studies, standardized reporting, and clinical workflow integration.