Non-Invasive Way for Detection of Neonatal Jaundice using GBR

Background: Neonatal jaundice is a common condition in newborns due to excess levels of bilirubin. The traditional method for bilirubin testing is invasive, i.e., through blood tests only, which is painful to infants. Therefore, this study uses machine learning algorithms to develop a non-invasive way to detect neonatal jaundice. Objectives: Design a computer-aided support system to detect neonatal jaundice using machine learning algorithm. Methods: The gradient-boosting regression model is used to predict the bilirubin level. Gradient Boosting is a robust boosting algorithm that combines several weak learners into strong learners, in which each new model is trained to minimize the loss function, such as mean squared error or cross-entropy of the previous model using gradient descent. Results: This study involves ensemble method which consist of non-linear methods that can improve the overall accuracy of the system. Conclusion: The hybrid approach for predicting bilirubin levels in neonates offers a promising non-invasive alternative to traditional blood tests. While the current model shows a positive correlation with ground truth values, further refinement is needed to address the observed discrepancies.