Applying a Random Forest Approach in Predicting Health Status in Patients with Carotid Artery Stenosis 30 Days Post...

Area of Science:

• Cardiovascular Medicine
• Neurology
• Health Services Research

Background:

• Carotid artery stenosis accounts for 20% of US ischemic strokes.
• Carotid artery stenting (CAS) reduces stroke risk, but post-procedure outcomes vary.
• Predictors of post-CAS health status require further understanding.

Purpose of the Study:

• To identify key predictors of 30-day health status after CAS.
• To develop predictive models for post-CAS health outcomes.

Main Methods:

• Utilized data from the SAPPHIRE Worldwide Registry for transfemoral CAS patients.
• Employed Random Forest models to rank 66 preprocedural variables by importance.
• Developed multivariable linear regression models to predict 30-day health status.

Main Results:

• Preprocedural health status, Modified Rankin Scale, and NIH Stroke Scale were key predictors.
• Medical comorbidities (CHF, diabetes, COPD) and ethnicity also influenced outcomes.
• Predictive models explained 36%-61% of health status variance.

Conclusions:

• Multivariable models partially explain 30-day post-CAS health status.
• Preprocedural status, stroke scores, and comorbidities aid in risk stratification.
• These factors are crucial for shared decision-making in CAS treatment planning.