A novel algorithm for fast BCG cycle extraction in ambulatory scenarios

Abstract

A novel algorithm JDet for ballistocardiogram (BCG) cycle extraction is presented that is intended for the fast generation of ensemble averages in ambulatory scenarios where the ECG signal is not available. First, the sensitivity (Se) and positive predictivity (+P) of JDet are evaluated on recordings obtained from 14 healthy subjects that stood on a modified weighing scale, and the results are compared with those from BSeg++, a popular BCG cycle-extraction algorithm. Second, the signal-to-noise ratio (SNR) is calculated for ensemble averages generated by using JDet, BSeg++, and the ECG R-waves on recordings of different duration. The results show that JDet has higher Se and similar +P than BSeg++. Therefore, the quality of the ensemble averages generated from JDet can be comparable to those generated from the ECG, the only cost being a 25% longer recording time.