Abstract
Neonates in neonatal intensive care units (NICUs) require continuous respiratory monitoring to manage high risks of apnea of prematurity and respiratory distress syndrome. Conventional monitoring techniques, such as transthoracic impedance pneumography and adhesive electrocardiogram leads, present significant clinical challenges, including epidermal stripping, elevated infection risks, and motion-induced false alarms. In this study, we present the design, implementation, and experimental validation of a miniaturized 24 GHz continuous-wave (CW) Doppler radar sensor system engineered specifically for non-invasive, continuous respiratory rate monitoring and automated apnea event detection in neonates. The hardware integrates an optimized microstrip patch antenna array with a high-sensitivity quadrature transceiver and low-noise baseband conditioning circuitry. To resolve the minute chest-wall displacements characteristic of preterm infants (<0.5 mm), we developed an adaptive signal processing pipeline incorporating complex demodulation, ensemble empirical mode decomposition, and a dynamic peak-detection thresholding algorithm. Benchtop phantom testing and simulated clinical validations demonstrate that the radar system achieves a root mean square error (RMSE) of 1.14 breaths per minute across a physiological range of 20–90 breaths per minute compared to standard clinical references. Furthermore, the automated apnea detection algorithm achieved a sensitivity of 97.4% and a specificity of 95.8% for cessation events exceeding 10 seconds. These findings highlight the potential of wearable CW radar as a clinically viable, contact-free modality to improve neonatal patient safety and comfort.
