Supported Research

Lay Abstract

Peripheral arterial disease (PAD) is a progressive condition in which narrowed or blocked blood vessels result in reduced blood flow to the legs. PAD results in impaired mobility, low quality of life, and other serious complications, such as foot ulceration and amputation. People with hemodialysis-dependent kidney disease develop PAD much more often than people with normal kidney function and also tend to develop more complications. Despite the burden of PAD, people with kidney disease are under-diagnosed and under-treated because we do not know how to detect it at earlier stages, before symptoms and complications develop, so this condition is usually only detected late, when mobility is affected or there is damage to the tissues. As detection and management prior to the onset of symptoms and complications may prevent serious complications, ways to identify PAD earlier would be of value to people living with dialysis dependence, clinicians, and policymakers. Using a person's health characteristics to predict their risk of developing a health condition in the future is an individualized approach to disease screening and diagnosis; yet, tools to predict PAD have not been used in people requiring hemodialysis. In this study, we will test if we can accurately predict individuals' risk of developing PAD after starting dialysis as well as their risk of PAD complications (foot ulcers or amputation). To do this, we will use data from a large clinical hemodialysis database, The Chronic Kidney Disease Cohort Study along with data we have on the actual number of times PAD and its complications occurred over time in these participants. We will determine how accurate our models are in predicting these actual events using a type of artificial intelligence, machine learning, and compare this to the accuracy of standard statistical models. An accurate PAD prediction model can then be used to improve access to optimal care through timely, early detection as the prediction tool can be used to identify who may benefit from early and more intensive monitoring prior to the development of complications. Identifying individual PAD risk could also be used in the design of clinical research, by assigning those at highest risk to potential preventative treatments.