Feature Selection of Photoplethysmograph Data in Machine Learning

Photoplethysmography signals are more responsive to changes in blood volume, not vascular pressure. Nowadays, more and more research is being developed for medical purposes, one of which is to diagnose diseases through fingertip pulse waves. This study proposes a new approach to optimize the statistical parameters of regression produced by PPG signals. The fingertip pulse wave device samples the PPG signal in humans and obtains the value of the signal. By taking the following samples, through processing using machine learning to process PPG signal data. machine learning is built to process PPG signal parameter data by the proposed method. The machine learning of feature selection algorithm that used are Forward Feature Selection Algorithm (FFS) and Sequential Input Selection Algorithm (SISAL). And for Machine learning using several methods is expected to obtain processing from the statistical parameters generated by Random Forest Regressor after going through feature selection by feature extraction with accuracy for FFS is 90% and for SISAL is 89%