Abstract 4368298: AI-Enhanced Automated Vascular Risk Assessment Using Multi-Territory Non-Coronary Imaging: A Systematic Review

Introduction: Standard cardiovascular testing relies primarily on medical history and coronary imaging, which may fail to detect the full extent of systemic vascular disease. Using AI in non-coronary vascular blood vessel areas, such as the carotid, aortic, and peripheral systems, may offer a thorough investigation and provide a better understanding of one's risk. Currently, manual assessments require a significant amount of time, and the varying interpretations of different experts often influence their accuracy. Hypothesis: Applying AI to the automated analysis of imaging for vascular diseases across multiple territories yields more accurate predictions of major adverse cardiovascular events than traditional risk scores and image analysis from a single area. Methods: Using PRISMA standards, we searched a range of databases (from 2018 to 2024) for studies focused on AI algorithms in non-coronary vascular imaging. Two researchers evaluated the studies and pulled out the information on predicting cardiovascular death, heart attack and stroke. Secondary outcomes included measuring the frequency with which each test produces the same result, assessing the time required for processing, and evaluating how well they correlate with established vascular biomarkers. A review of study quality was conducted using the QUADAS-2 and NOS. Results: All 23 selected studies (n = 28,894) found that AI-enhanced analysis of vascular parameters consistently predicted risk more accurately than traditional risk scales. Measurements of carotid thickness and stiffness of the aorta and arteries of the legs were highly reproducible and required much less analysis time using automation. Authors across various studies found that models using data from multiple areas performed better than models trained solely on one area. Conclusion: This analysis demonstrates that AI features in the automated analysis of imaging from multiple locations outside the heart enhance both the accuracy and efficiency of cardiovascular risk estimation. It enables thorough surveying of the vascular system for precision cardiology.