Cardio-respiratory Health Monitoring Using a Wearable Radar System

Abstract

Assessing cardio-pulmonary health is a crucial aspect of evaluating overall well-being. With the rise of cardiovascular and respiratory diseases worldwide, monitoring these physiological functions is vital for proper healthcare assessment. Cardiovascular and respiratory diseases pose significant global health concerns. According to the World Health Organization (WHO), cardiovascular disease (CVD) stands as the leading cause of global deaths, accounting for approximately 17.9 million deaths annually. Similarly, respiratory diseases such as chronic obstructive pulmonary disease (COPD) and asthma affect millions of people, resulting in high healthcare costs and a diminished quality of life. Monitoring cardiopulmonary health is of utmost importance, particularly in the wake of the COVID-19 pandemic. Studies reveal decreased cardiovascular and pulmonary performance post-infection, particularly among older adults and those with comorbidities. A traditional gold standard method for cardiac or respiratory monitoring, such as an electrocardiogram (ECG) or spirometry, has limitations. These limitations include complexity, skin irritations, lack of continuous monitoring, and the requirement for patient cooperation. Contactless radar-based monitoring techniques that use continuous wave (CW) and frequency-modulated CW (FMCW) radars offer a solution. These techniques provide detailed information on cardio-respiratory mechanics and rates by capturing the minor chest vibrations caused by respiration and cardiac activity, which are imperceptible to the naked eye. This sensitivity to minute vibrations is particularly pronounced in the case of millimeter-wave (mm-Wave) radar systems. FMCW radars come with advantages such as a simple architecture, customizable resolution, distance and velocity data, and continuous signal acquisition. However, these contactless far-field operation methods encounter challenges. These challenges prevent the widespread adoption of the technology and include random body movements, multiple targets, noise distortion, a limited field of view, and difficulties in accurately extracting cardiac information when simultaneous respiration is present. An FMCW radar operating in the near-field is proposed for use as a chest wearable, employing a mm-Wave radar module (Infineon, BGT60TR13C). This chest-wearable radar module functions at 60 GHz, enabling the uninterrupted extraction of various vital signs (VS) linked to cardio-respiratory activity. This system provides valuable information about the displacement waveforms associated with the movement of the chest during respiration and the cardiac cycle. Additionally, it accurately measures the respiratory rate (RR), heart rate (HR), and heart rate variability (HRV). To ensure the effectiveness of the radar antenna when in close proximity to the skin, comprehensive full-wave electromagnetic simulations were conducted. These simulations aim to evaluate the performance of the radar system and its ability to perform satisfactorily after attenuation due to the presence of the human torso on the near field. The results showed that the radar antenna demonstrates acceptable levels of resultant loading with skin in the near field, allowing for precise measurements of cardio-respiratory displacement waveforms. Furthermore, experiments to extract cardiac and respiratory waveforms and rates are conducted with the chest-worn prototype, with feature-rich cardio-respiratory displacement waveforms being successfully extracted. Extracted cardiac waveforms are detailed, and cardiac events are successfully mapped to a time-synchronized ECG signal acquired by a health monitoring belt (Fourth Frontier, Frontier X). Similarly, the HR and HRV were extracted with accuracies exceeding 97\% when compared to the Frontier X under breath-holding conditions. The effect of respiration on the quality of extracted cardiac signals is demonstrated, as it reduced accuracy of extracted cardiac biometrics to around 89\%. Moreover, the low sensitivity of the extracted cardiac signals and biometrics to the position of the radar on the chest is highlighted as the radar measures mechanical vibrations that propagate anywhere on the chest. Likewise, detailed respiratory waveforms are extracted that enable accurate estimation of RR, with an accuracy of 98\% achieved for the stationary RR estimation when compared to a ground truth metronome. We also demonstrate system effectiveness at differentiating between different types of breathing (laboured, shallow), and apnea with a high true positive rate by employing a statistical thresholding method. Finally, we explore the efficacy of the system as a respiratory monitoring tool during exercise and demonstrate its superior performance when compared to an off-the-shelf health monitoring device (i.e., Frontier, X) with time-synchronized RR data when compared to a ground truth metronome that regulates participants' breathing. To conclude, recommendations are made to increase system data security, throughput, and scalability by employing a custom power over Ethernet (PoE) system. The system diagram and the firmware implementation are summarized to highlight the interactions between different system components. Additionally, a recommendation is made to investigate a dual-band system that makes use of an additional frequency, particularly the 2.4-2.5 GHz frequency band as it offers better penetration through the human body, which could offer additional functional cario-respiratory information.