Vascular Age Evaluation Enhanced using Recurrence Plot Analysis and Convolutional Neural Networks

Abstract

Aging contributes as a major nonreversible risk factor for cardiovascular disease. This underscores the emergence of Vascular Age (VA) as a promising alternative metric to evaluate an individual’s cardiovascular risk and overall health. This study explores the use of a Convolutional Neural Network to estimate the VA group, as a surrogate of chronological age, utilizing Recurrence Plot as a robust tool for feature enhancement and image visualization from the Arterial Pulse Waveform (APW). The APW was obtained from an in-silico database of a one-dimensional cardiovascular model. The CNN demonstrated a robust performance, achieving an accuracy of 83% and 81.3%, an F1-score of 83.3% and 81.7% and an AUC of 0.96 and 0.95 for training and testing respectively. These findings may have potential implications for clinical applications, offering a non-invasive approach to cardiovascular risk assessment. The results contribute to the ongoing dialogue in cardiovascular research, highlighting the potential for innovative methodologies to enhance patient care and health outcomes. Further research will be essential to validate these methods for applications in real-world healthcare scenarios.