Fetal Health Classification using AI from Cardiotocography Features

Maternal health issues during pregnancy are a major global concern because they result in fetal death, which is more frequent in rising and poor countries. Cardiotocography (CTG) is the major non-invasive and cost-effective method for monitoring fetal health. CTG delivers critical parameters such as Fetal Heart Rate (FHR) and Uterine Contraction (UC), with a total of 21 features for accurate monitoring. These characteristics assist obstetricians in categorizing fetal health as normal, suspect, or pathological. However, differences in CTG interpretation among obstetricians frequently result in incorrect procedures. This work presents a unique way to detect deficient fetal states using a hybrid Machine Learning (ML) model that combines Support Vector Machine (SVM) and Random Forest (RF). The UCI repository dataset is pre-processed, which includes balancing, outlier reduction, and feature selection. The hybrid model’s performance is then compared to that of classic ML models such as SVM, K-Nearest Neighbour (KNN), RF, and Decision Tree (DT). The hybrid model outperforms the other models in simulation, attaining the greatest accuracy rate of 98.83%. This novel ML-based technique has the potential to improve fetal health detection while reducing the influence of human CTG interpretation discrepancies across healthcare providers.