Abstract: Arteriovenous fistula (AVF) stenosis is a leading cause of vascular access dysfunction in hemodialysis patients. Early and accurate diagnosis is crucial for reducing thrombotic risk and prolonging access lifespan. Traditional diagnostic methods have the limitation of invasiveness and the dependence of specific equipment. Machine learning enables non-invasive and objective diagnosis of AVF stenosis through extracting and quantifying acoustic features of normal and abnormal blood flow signals to establish recognition criteria. This article integrates machine learning with other four diagnostic tools to explain their principles, advantages and challenges, contributing to the shift toward precision medicine in vascular access management.

Key words: Arteriovenous fistula stenosis, Machine learning, Diagnostic tool, Hemodialysis