Cardiotocography (CTG) is widely used to monitor fetal heart rate (FHR) during labor and assess the wellbeing of the baby. CTG signals are commonly interpreted visually, challenging, mundane, and prone to error due to high inter- and intra-operator variabilities. While computer-based methods have been developed to detect abnormal CTG patterns automatically by mimicking clinical guidelines, they have poor accuracy due to a variety of complex reasons, resulting in missed opportunities to prevent harm as well as leading to unnecessary interventions. More recently, data-driven approaches using deep learning methods have shown promising performance in CTG classification to detect academia around the time of birth.
Our study utilises routinely collected CTGs from 51,449 births at term to classify births with and without severe compromise from the first 20 minutes of FHR recordings using deep learning techniques. We aim to detect abnormal CTGs as early as possible, preferably around the onset of labor, to allow adequate clinical decision and intervention time. I will talk about our methods, results, and future work directions.