Predicting Fetal Health: A Machine Learning Approach using Random Forest Algorithm

Abstract

In the existing literature, diverse methodologies, including cardiotocography, are employed to monitor pregnancies, encompassing the evaluation of heart rate, accelerations, fetal movements, and uterine contractions. This study proposes the utilization of advanced algorithms, specifically focusing on the Random Forest method which is used to classify illnesses. Confusion matrices are applied to discern normal and suspicious pathological states. The overarching objective of this research is the development of a robust predictive model for fetal health status by leveraging a spectrum of prenatal features. The primary goal is to establish a predictive model characterized by precision and reliability, facilitating the early detection of potential fetal health issues. Demonstrating a commendable accuracy score of 0.94, the random forest model emerges as highly effective in predicting fetal health based on a dataset featuring crucial parameters such as baseline values, accelerations, fetal movement, and uterine contractions. The purpose of this study is to significantly make a contribution to the discipline of prenatal healthcare by providing a sophisticated and precise tool for proactive fetal health assessment.