Abstract:
This work presents a non-invasive method for estimating Pulse Wave Velocity (PWV) from PPG signals using visibility graph transformation and machine learning. Complex features are extracted and selected through a multi-criteria approach. The study is based on simulated data (In-Silico) and real clinical data (VitalDB). The combination of different types of features allows a more comprehensive representation of the PPG signal structure. The results demonstrate the method’s ability to estimate arterial stiffness accurately and robustly.
