Edge-based Real-time Fetal Electrocardiography Monitoring in the Home Setting

Floranne Ellington, Berken Demirel, Daniel Jilani, Mohammad Al Faruque, Hung Cao
University of California, Irvine


Abstract

Health monitoring is increasingly moving to continuous, home-based applications with real time analytics. Fetal monitoring has been traditionally carried out in the hospital during checkups. While newly developed wearable systems offer home monitoring capabilities, rigorous real-time analytics require mobile or cloud connections for complex computation. However, these architectures result in higher energy output from the sensors and devices, reliance on a steady Bluetooth or WiFi signal, potential latency and synchronization variables, and data privacy issues. Furthermore, it is impossible to achieve telemedicine in resource-poor locations where mobile phone and internet are not accessible. To address these problems, we develop an edge computing paradigm for future integration in a compact abdominal patch to monitor the fetus remotely in the home. The energy saving and cost-effective algorithm, namely Lullaby, is developed to extract the fetal heartrate from a pregnant mother’s abdominal electrocardiography (ECG) signal. The algorithm was validated using the Physionet 2013 Challenge Dataset and achieved an average F1-score of 90%. Our experiments have shown that the proposed approach is deployable in a multitude of physical systems, either as a standalone fetal ECG monitoring patch or integrated into a more complex architecture. Additionally, evaluation on real hardware shows that our methodology is suitable for devices having a minimum RAM of 64KB which can be implemented on low-cost MCUs and designed to be energy saving for longer, continuous monitoring.