PREDICTING THE MODE OF CHILD DELIVERY THROUGH DATA MINING

Abstract

In maternity care, the mode of child delivery is crucial to prepare the patient mind considering her health condition before her due date. It helps obstetricians in decision-making to predict the most accurate childbirth mode in advance to avoid unnecessary emergencies and increase the survival of the mother and her unborn child. The type of delivery currently depends on the attending physician's judgment. Humans are prone to error; in several cases, doctors have recommended cesarean section birth when it is medically unnecessary. A wrong decision might also imperil the life of the mother and her unborn child. A computer-aided decision-making model could be a great solution to this problem. With this in mind, we used five data mining methods (Naive Bayes, Logistic model tree (LMT), Decision stump, Logistic Regression, and Random Forest) to create a model that predicted the best mode of infant birth. After training and analyzing the performance of each classifier using WEKA (Waikato Environment for Knowledge Analysis), our results showed that the Naives Bayes classifier has the highest accuracy, precision, and recall scores (89% each) and is the most stable model. This algorithm (Naives Bayes) is embedded to create a predictor that predicts the method of infant birth using a fresh data set that follows the trend of the previous data set. Using this model to decide the mode of childbirth will dramatically reduce maternal and newborn health risks.