Classification of Maternal Emergencies Using Gaussian Naive Bayes to Speed up the Patient’s Triage Process

Document Type : Original Article

Authors

1 Gadjah Mada University

2 Department of Electrical Engineering and Information Technology, Faculty of Engineering of Gadjah Mada University

3 Department of Obstetrics and Gynecology Faculty of Medicine, Public Health, and Nursing of Gadjah Mada University

Abstract

Introduction: Labor is the most important process in every woman’s pregnancy. This requires
optimal handling of various parties until labor takes place smoothly. The purpose of the study
is to determine the triage classification of labor referral patients in hospitals using Gaussian
Naive Bayes as the final model.
Methods: This study used 90 data, each consisting of 15 parameters which are divided into
two categorical data types: 9 data and 6 continuous data types. Two treatments were used in
this study, namely Gaussian Naive Bayes (first) using the independence assumption on all
parameters, and Categorical Naive Bayes for categorical data types, and Gaussian Naive Bayes
for continuous data types. These two types of data were combined using Gaussian Naive Bayes
as the final model. The data went through a preprocessing stage, stratified cross-validation;
then, we used the method of Naive Bayes according to the data type and continued for the
final stage classification using Gaussian Naive Bayes.
Results: The results of the first treatment had an accuracy of 91%, recall of 97%, precision of
64%, and F1-score of 73%. Also, the second treatment had an accuracy of 96%, recall of 88%,
precision of 86% and F1-score of 86%. The treatment of different data types had a difference
in the final results compared to the treatment of the same data type.
Conclusion: The diversity of data types is best handled according to the model used.

Keywords

Main Subjects