Prediction System Design for Monitoring the Health of Developing Infants from Cardiotocography Using Statistical Machine Learning

Abstract

Today segregation models are widely used in health care, which is intended to support physicians in diagnosing diseases and reducing human error. The challenge is to use effective methods to extract real-world data from the medical field, as many different models have been proposed with varying results. Many researchers have focused on the problem of variability in real-time data sets in segmentation models. Some previous works create mechanisms that include similar graphs for information display and information acquisition. However, such methods are weak in finding different relationships between elements. The purpose of this diagnostic method is to measure the baby's heart rate and uterine contractions during the third trimester of pregnancy, when the baby's heart is fully functional. Cardiotocogram findings are usually divided into three categories: physical, suspicious, or pathological. The purpose of this work is to automatically distinguish these regions using cardiotocographic data. In this study, the Random Forest method shows that it performs very well, capable of analyzing data with 94% accuracy. Comparisons with the Separation and Reversal Tree as well as mapping methods are also provided in the corresponding research paper.