Prediction of Pregnancy Complications in Fetal Heart Rate Using Hybrid Supervised Learning Model Enhanced With Compressive Sensing

Abstract

Complications during pregnancy are now common for most of the maternal. Those complications will affect both the maternal's and fetal's health. Women might have pre-pregnancy issues and it will prolong to their difficulties, others may subject to health relate issues during pregnancy. More crucially, fetal arrhythmias include tachycardia(high heart rate) and bradycardia(slow heart rate) may give rise to fetal heart damage. Normally, heart rates of fetal vary from 110 to 165 beats per minute. It seems to be abnormal, if it exceeds the distinctive ranges or if the rhythm is erratic. Thus, fetal heart rate monitoring is essential in anticipating pregnancy complications. For classifying pregnancy complications, a 2D Convolutional neural network (CNN) will be used at the data acquisition step, and it will be augmented using compressed sensing and modified orthogonal matching pursuit (MOMP).The reconstruction error, probability of detection, accuracy of detection will be used to quantify proposed techniques and the simulation results shows its superiority over unsupervised algorithm.