Artificial Intelligence and Machine Learning applications in Hemodialysis Arteriovenous Fistula

Machine learning and deep learning have emerged as powerful approaches for pattern recognition in complex physiological signals, offering the prospect of objective, automated, and continuously available vascular access monitoring.In this narrative review, we synthesise sixteen studies published between 2019 and 2026 that applied machine learning or deep learning to clinically relevant vascular access problems, including maturation prediction, stenosis and occlusion detection, aneurysm grading, access quality classification, and ultrasound image segmentation.The methods evaluated encompass artificial neural networks, gradient-boosted ensemble models (XGBoost, LightGBM), convolutional neural networks applied to photoplethysmography, acoustic and image data, vision transformers, and hybrid architectures.