Waves of Wellbeing - An Exploration of Remote Health Monitoring Across Species Using FMCW Radar
Ehsan Sadeghi is a PhD student in the Department of Pervasive Systems. (Co)Promotors are prof.dr.ir. M.R. van Steen, prof.dr. P.J.M. Havinga† and dr.ir. A. Chiumento from the Faculty of Electrical Engineering, Mathematics and Computer Science.
The increasing demand for advanced health monitoring systems in both human and veterinary fields has exposed critical limitations in traditional methods, particularly those requiring direct contact with sensors. While wearable devices have demonstrated effectiveness, they often face significant challenges, such as discomfort, inconsistent usage, and the stress they can induce in vulnerable populations, including the elderly, chronically ill patients, and newborn animals. These challenges underline the urgent need for innovative, non-invasive solutions. This thesis investigates the potential of Frequency Modulated Continuous Wave (FMCW) radar as a transformative technology for remote health monitoring, offering a non-invasive, accurate, and privacy-sensitive alternative that overcomes the limitations of traditional approaches.
The research begins by identifying key health indicators that are essential for early detection and effective management of health issues. These include vital signs, such as heart rate and respiratory rate, and behavioral patterns, including activity and posture.
Chapter 2 provides a comprehensive analysis of the most pressing health challenges and mortality causes in humans, livestock, and pets, emphasizing the critical role of these indicators in proactive and preventive healthcare. By linking these indicators to specific health conditions, the groundwork is laid for the exploration of sensing technologies capable of monitoring these parameters continuously and reliably.
Chapters 3 and 4 evaluate a broad range of sensing technologies, comparing wearable and remote sensing approaches to determine their suitability for monitoring the identified indicators. While wearable devices are often limited by their intrusive nature and reliance on user compliance, remote sensing technologies—such as cameras, thermal imaging, and Wi-Fi-based systems—offer significant advantages. Among these,FMCW radar emerges as the most promising due to its high resolution, adaptability to diverse environments, and ability to ensure privacy. These chapters also introduce the principles of FMCW radar technology and outline its capability to monitor health indicators with precision, establishing it as a robust solution for both human and animal applications.
The application of FMCW radar to human health monitoring is explored in detail in Chapters 5 and 6. These chapters present the development and validation of advanced signal processing techniques and AI-driven methodologies to enhance the radar’s ability to detect vital signs and recognize behavioral patterns in real-time. Specifically, novel algorithms are introduced for vital sign estimation, achieving high accuracy across a variety of scenarios, including extreme physiological conditions. For activity and posture recognition, the study employs two approaches: using micro-Doppler signatures to capture motion dynamics and generating point cloud data for spatial analysis. A comparative evaluation reveals that combining deep learning techniques, such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), significantly enhances classification performance. These findings underscore FMCW radar’s potential as a reliable, non-invasive, and privacy-preserving tool for continuous human health monitoring. Building on these findings, Chapters 7, 8, and 9 extend the application of FMCW radar to veterinary health monitoring, addressing the unique challenges associated with animals. Monitoring animals involves complexities such as size variations, random movements, and environmental factors specific to farm and domestic settings. To adapt the methods developed for humans, the thesis refines signal processing pipelines and customizes AI models to account for species-specific physiological and behavioral traits. Experiments conducted with calves, dogs, and cats demonstrate the feasibility of these adaptations. For example, vital sign detection in calves requires overcoming challenges posed by their small size and constant movement, while activity and posture recognition in dogs and cats necessitate advanced preprocessing and machine learning techniques to handle their unique motion patterns. Despite these challenges, the results validate the effectiveness of FMCW radar in delivering accurate, non-invasive health monitoring for animals, highlighting its potential to significantly enhance welfare standards and farm productivity.
The concluding chapter, Chapter 10, synthesizes the findings and contributions of this research. It reflects on the successful demonstration of FMCW radar as a versatile tool for remote health monitoring, capable of addressing both human and veterinary needs. The thesis highlights its advancements in signal processing and AI methodologies, which have enabled robust, real-time monitoring across diverse scenarios. The broader implications of this work include the potential to bridge gaps in healthcare access, enhance the quality of life for humans and animals, and establish a foundation for future innovations in remote sensing technologies. In addition, the appendix (Chapters 11 and 12) provides supplementary materials, including detailed datasets and case studies that offer deeper insights into specific applications, such as piglet health monitoring.
The structure of the thesis reflects a logical progression, with each part building upon the previous one to create a cohesive narrative. It begins by establishing the importance of health indicators and the limitations of existing monitoring systems. The subsequent evaluation of sensing technologies identifies FMCW radar as the most viable option, leading to the development of innovative methodologies for human applications. These methodologies are then adapted to veterinary contexts, addressing species-specific challenges and validating the radar’s versatility. The interconnections between chapters emphasize the adaptability of FMCW radar, demonstrating how techniques developed for humans can be effectively translated to animals, ensuring broader applicability across species.
By bridging the gap between advanced radar technologies and practical healthcare needs, this research makes significant contributions to biomedical engineering and veterinary science. It establishes FMCW radar as a pivotal technology for non-invasive, privacy-sensitive health monitoring, setting new standards for healthcare innovation.
This thesis not only addresses existing challenges in health monitoring but also lays the groundwork for future advancements, paving the way for more inclusive, efficient, and humane healthcare solutions across species.
