ABSTRACT Background Continuous vital sign monitoring with wearable biosensors enables earlier detection of clinical deterioration and improves patient safety. This study validated an integrated wireless photoplethysmography‐based chest patch biosensor (Biobeat technology Ltd) capable of measuring five vital signs relevant for Early Warning Score calculation, by comparison with standard continuous bedside monitoring in the intensive care unit. Methods Agreement between the wearable biosensor and the reference monitor was assessed using Bland–Altman analysis for repeated measurements. Bias and limits of agreement were calculated for all vital signs. Measurement differences were compared with predefined accuracy limits. Data were aligned at 5‐ and 15‐min intervals using median reference values. Deming regression, Pearson correlation, and Clarke Error Grid analyses were performed. Results‐ A total of 1931.8 monitoring hours were collected from 32 intensive care patients (mean 60.4 ± 28.5 h). Heart rate showed a bias of 0.16 beats per minute (LoA −17.06 to 17.38), with 18.8% exceeding the 10% threshold and 0.42% in Clarke Error Grid zones D–E. Respiratory rate showed a bias of 3.77 breaths per minute (LoA −6.78 to 14.31), with 61.4% exceeding the 10% threshold and 18.6% in zones D–E. Systolic blood pressure, oxygen saturation, and temperature showed biases of −4.69 mmHg, −1.34%, and −0.29°C, respectively. Conclusion Agreement between a wearable chest‐patch biosensor and standard bedside monitoring was evaluated in intensive care patients. Overall agreement was variable, with wide limits of agreement and a substantial proportion of measurements outside predefined thresholds. Further optimisation and validation are required before broader clinical application. Editorial Comment In this article, limitations in agreement between signals from wearable wireless sensors for vital signs and usual vital signs monitor values in intensive care cases are described for data collected from a practical and pragmatic setting where the study participants were in early phase of their illness with some expected clinical deterioration. There were enough test device values outside of predefined limits of agreement based on clinically relevant differences, that the authors interpret this as a need for further improvement in device performance.
Slambrouck et al. (Sun,) studied this question.