Real-time pre-eclampsia prediction model based on IoT and machine learning

Abstract

Abstract Pre-eclampsia (PET) is a hypertensive disease that occurs during pregnancy or in the postpartum period. It complicates 2% to 8% of all pregnancies and is one of the causes of more than 50,000 maternal deaths and over 500,000 fetal deaths worldwide annually. Adverse birth outcomes due to pregnancy complications have been associated with three delays: delay in recognizing the complication, delay in reaching an appropriate facility, and delay in receiving adequate care when the facility is reached. Thus prevention, timely detection, and care of pregnancy complications can prevent maternal deaths and morbidity. The Internet of Things (IoT) and machine learning (ML) technologies have become the new revolution of research in the field of healthcare. These technologies can be utilized to interconnect various sensors, monitor the health status of a patient, and predict the occurrence of an ailment. This study has designed and prototyped a pre-eclampsia monitoring model based on IoT and machine learning for remotely monitoring the health status of an expectant woman and her unborn child, to enhance early diagnosis of pre-eclampsia and improve birth outcomes. The study involved researching the on most appropriate biosensors and then designing and prototyping the pre-eclampsia watch. To build the pre-eclampsia prediction model the best ML algorithm was empirically analysed. A Naïve Bayes pre-eclampsia prediction model was found to perform better in identifying pregnant women who are at risk of pre-eclampsia after evaluation of various pre-eclampsia models built using decision trees, Naïve Bayes, K Nearest Neighbor (KNN), logistic regression, support vector machines (SVM) and Artificial neural networks (ANN). Lastly, the predictive model was integrated with the pre-eclampsia model to assist in early diagnosis of pre-eclampsia. The prototype generates alerts when the expectant woman is at risk of Pre-eclampsia. The pre-eclampsia watch model can securely capture and transmit expectant women's vital to the cloud for processing and provide timely alerts when the woman is at risk. Further research on the performance and efficacy of the model in a real environment will be done by experimenting with it in a purposively selected sample.

Keywords

• Pre-eclampsia
• Internet of Things
• Machine learning
• Pregnancy monitoring