Feature Engineering Analysis for Predicting Cardiovascular Aging: An In-Silico Approach

Studying non-invasive markers of Cardiovascular Health (CVH) is crucial, considering cardiovascular diseases remain one of the leading causes of mortality worldwide. This study comprehensively explores Cardiovascular Age (CVA) using photoplethysmographic waveforms (PPGW) from an in-silico dataset, highlighting the importance of feature selection in supervised learning algorithms. Through rigorous analysis, robust biomarkers such as PTT and PPGms were identified as potential non-invasive markers for assessing CVH. This research establishes guidelines for optimal feature selection and sensor placement, laying a foundation for accurate CVA estimation using PPGW. Further work with in-vivo data should be performed. The present work contributes to advancing PPG-based CVA estimation and the development of non-invasive tools for CVH assessment.