CLASSIFICATION MODEL FOR CARDIOTOCOGRAM MACHINE USING FEED FORWARD NEURAL NETWORK

One of primary concerns during pregnancy is the challenge of accurate monitoring of Fetal Heart Rate (FHR) and interpretation of cardiogram data to help determine the condition of the baby. The aim of this study is to present a classification model for cardiotocogram machines using Feed-forward Neural Network (FNN). The dataset for this study was collected from the Kaggle repository. The data consist of 2126 fetal cardiotocogram (CTG) data with 42 attributes, span across three classes of normal, suspect and pathology. The data was processed using noise filtering, normalization and then splitted into training, testing, and validation set in the ratio of f 70:20:10. FFNN was then trained using the data and Levenberg-Marquardt as the optimization technique. The model generated was evaluated and the results reported 90.3% accuracy, 81.9% sensitivity, 82% specificity, and 80.9% precision. Comparative analysis with other state of the art algorithm was performed, with the results showing the competence of the new cardiotocogram classification model among the best. The model was recommended to improve operation and functionality of CTC machine.