Software Platform for CTG Classification Using Artificial Intelligence

• Published in: 2023 Seventh International Conference on Advances in Biomedical Engineering (ICABME) pp. 244 - 250
• Main Authors: Christelle, Margossian, Sandy, Rihana
• Format: Conference Proceeding
• Language: English
• Published: IEEE 12.10.2023
• Subjects: Cardiotocography; Feature extraction; Machine learning; Machine learning algorithms; MATLAB; Medical services; Pre-term Labor; Pregnancy; Signal processing algorithms; Software algorithms; Uterine contractions
• ISSN: 2377-5696
• DOI: 10.1109/ICABME59496.2023.10293080

Pregnancy is a critical period in every woman's life, and it comes with its own risks and downsides that she must look out for, such as pre-term labor. Unfortunately, maternal and fetal morbidity and mortality have been significant issues in the past decades that have yet to be resolved. Despite the evolving medicine and technology, tackling the dangers of pregnancy is still a challenge even though there are a lot of new devices used for monitoring. The target of this project is to build an algorithm that allows the signal processing and feature extraction of uterine contraction signals outputted from the Cardiotocography device, as well as a machine learning algorithm that classifies these signals into pathological, suspicious, and normal cases. These signals and features were collected from a trusted database and used to undergo signal processing before executing the machine learning step. Several algorithms were tested in this work including KNN, CNN, LMNN, and BNN where the Bayesian model exhibited the highest performance results with the input data. The input was also manipulated using the Chi squared approach to yield ameliorated results and analyzed based on significance.