Design of an EHG based Smart Labour Detection System

The purpose of this paper is to develop a portable noninvasive system to measure the EHG recordings and to distinguish between true and false labour. This proposed system would help the medical personnel to achieve a quicker and clearer diagnosis, which will reduce medical complications. This will also help the patients monitor their uterine contractions without any medical assistance and seek medical attention at the appropriate time. The EHG signals represent uterine muscle contractions and can provide useful information. These signals are recorded using electrodes and sampled using the ADC unit in the microcontroller. The recorded signals are processed and analyzed using the machine learning model on the microcontroller. The algorithm decomposes the signals and extracts their corresponding features using a regression model. These features are then used to train a classifier, which is used to distinguish between true labour and false labour. The Icelandic 16-electrode Electrohysterogram Database was used as the dataset for training and testing the model. This algorithm managed to produce an accuracy of 97.78%, an F1 score of of 97.87% and a standard deviation of 3.5%.