| Abstrakt: |
Newborns may develop a common condition at the start of their lives known as neonatal jaundice. High levels of bilirubin in the infant's blood cause jaundice due to immature liver. Additionally, it may lead to severe symptoms and serious complications. Thus, early detection of this condition is mandatory to prevent further complications. Current methods for measuring bilirubin level involve collecting blood from the patient. However, invasive techniques are stressful and painful and may cause unwanted complications, especially, when dealing with uncooperative patients like neonates. In order to avoid invasive methods, researchers sought other non-invasive methods to diagnose jaundice using image processing techniques and machine learning algorithms. This study offers a comparative performance between six machine-learning algorithms (MLA), namely, Naïve Bayes, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree (DT), LightGBM, and Random Forest (RF), based on a dataset of normal and jaundiced infant images. The results show that the RF has the highest performance among other algorithms used in this study, with an accuracy of 97.37%. At the same time, Naïve Bayes has the lowest performance of 88.16%. The results confirm that the Random Forest algorithm has the upper hand in binary classifications that help detect neonatal jaundice. [ABSTRACT FROM AUTHOR] |
