Estimating of CAVI-Based Vascular Age using Photoplethysmography Waveform Feature

Abstract

This paper proposes a method to estimate the vascular age from the reflected photoplethysmogram (PPG) waveform. Mobile devices such as smartphones are rapidly penetrating daily life, and health management using them is expected. This paper deals with a method of estimating the vascular age using a reflective PPG sensor attached to a mobile device. In the proposed method, contour lines are drawn on the PPG and the derivative PPG waveform, and features are extracted from them. Since many features are extracted, the influences of multicollinearity are concerned. Therefore, partial least squares regression which is not affected by multicollinearity was used as an estimation method. We compared the estimation accuracy with the conventional method using height ratio feature.