OPTIMIZATION OF PARTICLE SWARM OPTIMIZATION IN NAÏVE BAYES FOR CAESAREAN BIRTH PREDICTION

Abstract

The Maternal Mortality Rate (MMR) in 2017 according to the World Health Organization 
(WHO) is estimated to reach 296,000 women who die during and after pregnancy or 
childbirth. Caesarean birth is the last alternative in labor if the mother cannot give birth 
normally due to certain indications with a high risk, both for the mother and the baby. factors 
of a mother giving birth by caesarean section, such as placenta previa, hypertension, breech 
baby, fetal distress, narrow hips, and can also experience bleeding in the mother before the 
delivery stage. It is hoped that delivery by caesarean method can minimize problems for the 
baby and mother. Accurate prediction of the condition of the mother's pregnancy can enable 
d 
octors, health care providers and mothers to make more informed decisions regarding the 
management of childbirth. To predict caesarean births, data mining techniques using the 
Naive Bayes algorithm can be used. Naive Bayes is very simple and efficient but very sensitive 
to features, therefore the selection of appropriate features is very necessary because 
irrelevant features can reduce the level of accuracy. Naive Bayes will work more effectively 
when combined with several attribute selection procedures such as Particle Swarm 
Optimization. In this study, the researcher proposes a Particle Swarm Optimization 
algorithm for attribute weighting in Naive Bayes so as to increase the accuracy of Caesarean 
birth prediction results