Predicting intravenous thrombolysis outcomes in acute ischemic stroke using machine learning

Area of Science:

• Neurology
• Medical Informatics
• Biostatistics

Background:

• Intravenous thrombolysis is crucial for acute ischemic stroke (AIS) treatment.
• Predicting post-thrombolysis outcomes is vital for personalized patient care and clinical decision-making.

Purpose of the Study:

• To develop and validate machine learning models for early prediction of functional outcomes in AIS patients post-thrombolysis.
• To identify key predictors of poor functional outcomes to guide individualized treatment strategies.

Main Methods:

• Retrospective analysis of 383 AIS patients treated with intravenous thrombolysis.
• Development of Random Forest (RF), Gradient Boosting Machine (GBM), and Support Vector Machine (SVM) models using selected features from LASSO regression.
• Performance evaluation using AUC, calibration curves, and decision curve analysis; interpretability assessed with SHAP values.

Main Results:

• Higher admission NIHSS, elevated blood glucose, higher 24-h NIHSS, increased infarct core volume, higher HbA1c, and elevated Neutrophil-to-Lymphocyte Ratio were risk factors for poor outcomes.
• Larger ischemic penumbra volume was a protective factor.
• The RF model achieved superior predictive performance (AUC=0.784) with better calibration and clinical utility.

Conclusions:

• The developed machine learning model effectively predicts functional outcomes in AIS patients after thrombolysis.
• Identified predictors offer insights for clinical prognosis and optimizing thrombolytic therapy strategies.