Artificial Intelligence Techniques for Prognostic and Diagnostic Assessments in Peripheral Artery Disease: A Scoping Review

Abstract

Peripheral artery disease (PAD) is a major public health concern worldwide, associated with high risk of mortality and morbidity related to cardiovascular and adverse limb events. Despite significant advances in both medical and interventional therapies, PAD often remains under-diagnosed, and the prognosis of patients can be difficult to predict. Artificial intelligence (AI) has brought a wide range of opportunities to improve the management of cardiovascular diseases, from advanced imaging analysis to machine-learning (ML)-based predictive models, and medical data management using natural language processing (NLP). The aim of this review is to summarize and discuss current techniques based on AI that have been proposed for the diagnosis and the evaluation of the prognosis in patients with PAD. The review focused on clinical studies that proposed AI-methods for the detection and the classification of PAD as well as studies that used AI-models to predict outcomes of patients. Through evaluation of study design, we discuss model choices including variability in dataset inputs, model complexity, interpretability, and challenges linked to performance metrics used. In the light of the results, we discuss potential interest for clinical decision support and highlight future directions for research and clinical practice.