Comparative Analysis of Different ML Models Towards Term and Pre-Term Classification

Prediction of preterm birth represents one of the biggest obstacles in maternal care. This manuscript utilizes EHG signals as an early warning tool in preprocessing by using Butterworth bandpass filtering to eliminate interference. The selected features to extract are Sample Entropy, Median Frequency, Peak-to-Peak Amplitude and Root Mean Square. Machine learning algorithms, from which Gradient Boosting, Decision Tree, DNN, Random Forest, KNN and SVM were used for classification. Gradient Boosting has shown the highest accuracy of 88.33 %. Such research opens up signals processing combined with machine learning for improvement in preterm birth prediction, which can lead to early intervention and better prenatal care for both mothers and newborns.