Development and validation of a machine learning model based on multi-source clinical data for predicting the risk of early neurological deterioration in patients with ischemic stroke - Report - MDSpire
Conexiant’s news site is now MDSpire News. Learn more
Advertisement
Creation and assessment of a machine learning framework utilizing diverse clinical data to forecast early neurological decline in individuals with ischemic stroke
Clinical Report: Machine Learning Framework for Predicting Early Neurological Decline
Overview
This study developed and validated a machine learning model to predict early neurological deterioration (END) in patients with acute ischemic stroke. The model demonstrated satisfactory predictive performance using five clinical variables.
Background
Early neurological deterioration (END) is a significant clinical event following acute ischemic stroke, associated with poor outcomes and increased mortality. Identifying patients at high risk for END is crucial for timely intervention and resource allocation. Machine learning offers a promising approach to enhance predictive accuracy using diverse clinical data.
Data Highlights
Variable
Value
Patients with END
161 (14.0%)
AUC (Development Cohort)
0.787 (95% CI: 0.735–0.839)
AUC (Validation Cohort)
0.751 (95% CI: 0.668–0.834)
Key Findings
END occurred in 14.0% of the studied patients.
Five principal predictors identified: ischemic stroke subtype, OCSP classification, age, atrial fibrillation history, and previous stroke history.
The logistic regression model showed AUCs of 0.787 and 0.751 for development and validation cohorts, respectively.
SHAP analysis indicated that ischemic stroke subtype and age were the most influential predictors.
Decision curve analysis indicated potential clinical utility for the logistic regression model.
Clinical Implications
The developed logistic regression model utilizes routinely available clinical variables.
Conclusion
The study successfully developed a logistic regression model for predicting END risk, demonstrating satisfactory performance and interpretability.
New findings from CREST-2 are providing important evidence on the role of carotid artery stenting (CAS) in patients with severe asymptomatic carotid stenosis — while underscoring the importance of operator experience, patient selection and procedural technique in achieving favorable outcomes.
