Non Invasive Wearable Device for Fetal Movement Detection

Abstract

Monitoring fetal movement patterns is a very common method of assessing fetal health. Currently, there is a lack of a proper device to identify and monitor fetal movement patterns. Therefore in this research, a wearable device with an INS sensor was designed and fabricated to monitor fetal movement. The time-domain data acquired from the device was fed into three analysis methods to separate the fetal movements from the data. Initially, a direct deep learning algorithm was applied. Then a hybrid method where a standard signal processing algorithm combined with CNN was applied. The direct deep learning algorithm identified fetal movements with an average accuracy of 73%. The hybrid method where STFT was combined with CNN identified fetal movement with an average accuracy of 88%.