Applications of Artificial Intelligence in Major Vascular Diseases: A Systematic Review of Artificial Intelligence-driven Diagnosis, Risk Prediction, and Decision-making in Aortic Aneurysms, Aortic Dissections, Carotid Artery Disease, and Peripheral Arterial Disease (2005–2025)

Background: Artificial intelligence (AI) has emerged as a promising tool for improving diagnosis, risk prediction, and clinical decision-making in vascular surgery. However, the evidence supporting AI applications across major vascular diseases remains fragmented, with variations in methodology, validation, and clinical implementation. This systematic review evaluates the current role of AI in aortic aneurysms, aortic dissections, carotid artery disease, and peripheral arterial disease (PAD). Methods: A systematic literature search was conducted across PubMed, Google Scholar, and Scopus for studies published between January 2005 and December 2025. Studies evaluating AI, machine learning, or deep learning (DL) applications in diagnosis, risk prediction, outcome forecasting, or clinical decision support for major vascular diseases were included in the study. Data on study design, AI methodology, clinical application, and performance metrics were extracted. Methodological quality was assessed using the ROBINS-I tool. PROSPERO CRD420251207656. Results: A total of 289 records were identified, of which 30 studies met the inclusion criteria. AI applications were most extensively studied in aortic aneurysms, aortic dissections, and carotid artery disease. In aortic aneurysms, AI systems increased detection rates by 38.8% and predicted postprocedural adverse events with area under the curve (AUC) values exceeding 0.96. In aortic dissections, multimodal DL models identified mesenteric malperfusion risk (AUC: 0.78) and predicted postthoracic endovascular aortic repair adverse events with AUCs approaching 0.99. For carotid artery disease, AI models detected symptomatic plaques with AUCs of 0.89–0.90, outperforming conventional stenosis-based assessments. Evidence for PAD was comparatively limited, with few studies reporting robust clinical performance metrics. Most included studies were retrospective cohorts with moderate-to-serious risk of bias. Conclusions: AI demonstrates substantial potential to enhance diagnosis, risk stratification, and procedural planning in major vascular diseases. The strongest evidence currently exists for aortic aneurysms, aortic dissections, and carotid artery disease, where AI consistently outperforms traditional approaches in several diagnostic and prognostic tasks. However, the predominance of retrospective studies and limited external validation highlights the need for prospective multicenter studies to establish clinical utility and support widespread implementation in vascular surgical practice. Systematic review registration: CRD420251207656 available at https://www.crd.york.ac.uk/.