For the hundreds of thousands of people who depend on hemodialysis to survive kidney failure, a small surgical connection between an artery and a vein in the arm — called an arteriovenous fistula, or AVF — is essentially a lifeline. When that fistula stops working, dialysis cannot happen. Complications from AVF dysfunction are a leading cause of hospitalizations among dialysis patients.
A systematic review published in the medical journal Cureus examines how machine learning and artificial intelligence tools are being applied to predict when these fistulas might fail — before the failure actually occurs.
The review surveys the landscape of AI-driven approaches currently being studied, asking which algorithms and data inputs show the most promise for flagging dysfunction early. Rather than presenting new experimental results, it synthesizes existing research to give clinicians and researchers a clearer picture of where the field stands.
Predicting fistula problems in advance could allow care teams to intervene earlier — scheduling repair procedures or adjusting treatment plans — rather than reacting to an emergency. That kind of proactive care could reduce hospitalizations, lower treatment costs, and improve outcomes for a patient population that already faces a heavy medical burden.
If AI-assisted monitoring proves reliable enough for routine clinical use, it could meaningfully change how nephrologists manage one of the most critical — and fragile — parts of kidney care.