Using Discriminant Analysis to Indicate the Most Important Effective Factors of Getting Neonatal Jaundice "A Study at Benghazi Children’s Hospital"

Abstract:

This study aims to evaluate the accuracy and effectiveness of a statistical model in predicting the incidence of neonatal jaundice. The research identifies the primary factors influencing the onset of jaundice by analyzing data from newborn infants. Additionally, the study determines the relative importance of the variables associated with the likelihood of jaundice in neonates and develops a discriminant function capable of classifying affected newborns into two categories: normal and abnormal. The study produced several important findings. Most of the variables demonstrated statistical significance at a level below 0.05, indicating that they play a critical role in distinguishing between the two groups, apart from gestational age, which was found to be statistically non-significant. Using discriminant analysis, the model correctly classified 83.7% of the normal cases and 90.0% of the abnormal cases. The overall correct classification rate for the sample was 87.7%, suggesting that the model successfully classified 87.7% of all observations. This result reflects a high level of model performance and provides strong evidence of its overall predictive efficiency.

Keywords:

Jaundice, Discriminant Analysis, Discriminant Function, Latent Root, Model Classification, Relative Importance.