Features extraction for blood pressure estimation with photoplethysmography by autoencoder

Abstract

Blood pressure estimation methods have been proposed on the basis of the physiological knowledge that the PPG changes depending on the state of the cardiovascular system. In previous studies, various features which are extracted the wave height and the elapsed time from the rising point of the pulse wave to feature points have been used to the machine learning for the BP estimation. However, the accuracy is still not adequate as medical equipment because their features cannot express fully information of the pulse waveform which changes according to the BP. And, no other effective knowledge about pulse waveform for BP estimation has been found yet. Therefore, in this study, we focus on the autoencoder which can extract complex features and add new features of the pulse wave form for the BP estimation.