Detecting High-Risk Pregnancies And Premature Births: A Comprehensive Survey

Abstract

High-risk pregnancy detection is important for maternal and fetal health in regions having limited access to medical resources. Our solution presents identification for resource-limited settings using Artificial neural networks ANNs for predicting high-risk pregnancies early and precisely. Which enables improved health outcomes and timely intervention.  ANN model design, implementation, evaluation, and addressing healthcare challenges with resource limitation. High-risk pregnancy may cause several issues including lifelong health disabilities. We aim to diminish the consequences of high-risk pregnancies and test systems for their reliability and accuracy. Expert ANN system and back propagation algorithm which shows results with a 0.98 accuracy rate. We utilized a dataset comprising 172 medical records from patients, featuring 17 input parameters, and encompassing 5 distinct output classes. These classes included normal early pregnancy as well as four categories denoting various pregnancy disorders. Through a rigorous training and testing process, our experiment demonstrated the feasibility of applying an Artificial Neural Network (ANN) to predict pregnancy disorders. Notably, our model achieved an accuracy rate of approximately 78.248\%. This achievement was attained through meticulous parameter tuning: a learning rate of 0.1, input layers with 17 neurons, 5 neurons at output layer, layers that are hidden with multiple neurons i.e., 50 and an error value of 0.01.