Maternity Risk Prediction Using IOT Module with Wearable Sensor and Deep Learning Based Feature Extraction and Classification Technique

Abstract

Pregnancy is term used to describe the period of development of a fetus in the uterus of a woman, which is a period of about 9 calendar months, 40 weeks, or 280 days. Detection of pregnancy risk at an early stage significantly increases the survival rate. This paper proposes the maternity risk prediction through IOT module based wearable sensor and prediction of abnormal data collected from the wearable sensor and pre-historic data using DL based feature extraction as well as classification techniques. Here attributes like age, weight, BMI, fetus heart rate, current month, medical history of mother has been considered. Overall, results of experiments show that performance of proposed technique increases. Because AUC is a statistic that considers the entire FPR range, using it as a stopping criterion for early halting could also prevent us from developing a model that provides the best TPR at 10% FPR.