A HYBRIDIZED ANALYTICAL MODEL FOR MATERNAL OUTCOME CLASSIFICATION

ABSTRACT

Computational methods have been very successfully applied to many medical domains in solving real-life problems. Predicting maternal outcomes is an added advantage for health and non-health workers. The goal of this work is to find out the existing algorithms that could aid maternal outcomes, possible factors that can be considered as major decision variables in designing maternal outcomes, and to develop a framework model for evaluating the performance of maternal outcomes. Two computational approaches; the hybridized machine learning approach where Random Forest (RF) and Naive Bayes (NB) were combined using ensemble method and intuitionistic type-2 fuzzy set was used. The intuitionistic type-2 fuzzy set method had an accuracy of 91% while the hybrid approach got an accuracy of 95%. However, when comparing the result we found that hybridized method obtained a high accuracy than that of the intuitionistic type-2 fuzzy set. Although, both methods could be implemented to assist mothers within the bearing age, and make it easy for the physicians to carry out their work without much stress, since the predicted accuracy was within the acceptable error rate.

Keywords: Maternal Health, Machine learning, Hybridization and intuitionistic type-2 fuzzy set