Machine Learning Applications in Maternal Health Monitoring

Machine learning plays a main role in monitoring maternal health by detecting risk factors, best maternal care and enhanced decision making in clinical status. Machine learning methods could focuses on maternal datasets such as demographic data, laboratory results, medical history to determine risk factors like hypertension and gestational diabetes. Machine learning algorithms reviewed in this research are neural network, boosting methods, random forest, K-nearest neighbor, decision trees and hybrid model. Articles were chosen from recent six years for extensive search. Total number of research studies searched in the repository were 278. After filtering the research papers were 86 duplicates. Screened papers for this research were 192. Some papers are not included with inappropriate content were 92. Full text papers included in this study were 100. The most used machine learning algorithms for monitoring maternal health are random forest, followed by boosting methods, decision tree, hybrid methods, neural network and k-nearest neighbor. It was evident from the deep search of literature that, no single method is effective in monitoring maternal health. Every algorithm has its own strengths and weaknesses. It was evident from the literature that hybrid models have the best balance between robustness and predictive accuracy and make them suitable option for monitoring in real time and evaluating risk factors during pregnancy and overall supports in health outcomes of the mother and fetus. In Future, we can extend the study by developing a hybrid model by combining machine learning algorithms to monitor maternal health.