Increasing the Accessibility of Peripheral Artery Disease Screening with Deep Learning

Abstract

Peripheral arterial disease (PAD) is estimated to affect 200 million people worldwide and is one of the leading causes of limb loss. The early diagnosis and treatment of PAD is crucial in preventing adverse clinical outcomes. However, the current point-of-care diagnostic evaluation for PAD, the ankle-brachial index (ABI), has significant limitations making it underutilized. The goal of this study is to develop a deep learning-enabled system which can predict clinically relevant ABI ranges directly from circulatory sounds of an artery derived from a hand-held doppler. Our IRB-approved clinical study focuses on ubiquitous, efficient, and secure collection of data for training of our system as well as a pipeline for continuous validation across multiple sites. We approach the collection of data and deployment of our AI through a mobile application which provides a simplistic and intuitive platform to be used by vascular technologists and clinicians. Our work can contribute a unique point-of-view into the deployment of AI for niche medical applications and the usage of a mobile application to streamline AI clinical studies.