The Accuracy of a Commercial Wearable Neonatal Home-Monitoring Device: A Simulation Study

Wearable newborn-monitoring devices are increasingly popular and widely available for home use. One approach includes the use of accelerometer detection of breathing movements. We investigated the accuracy of breathing and apnoea detection by a wearable accelerometer in a simulated environment.

This study used simulation as a method to evaluate the performance of a wearable accelerometer-based breathing detection device. A term mannequin fitted with an accelerometer suit wirelessly linked to a mobile app (Goldilocks) was NeoPuff ventilated through a cuffed endotracheal tube. Peak inspiratory pressure (PIP) levels of 5-25 cmH₂O were delivered with zero end expiratory pressure. Each trial consisted of 30 s of ventilation, followed by simulated apnoea (n = 150). App-based detection and alert times were recorded blinded to ventilation status. Detection accuracy across PIP levels was analyzed using Kruskal-Wallis tests followed by Dunn's multiple comparisons with Bonferroni correction.

PIP level affected both apnoea detection (H(4) = 58.9, p < 0.0001) and breathing detection (H(4) = 86.5, p < 0.001) with performance reduced at 5 cmH₂O compared to higher PIP. Apnoea and breathing detection were most accurate at 20 cmH₂O. Mean (SD) apnoea detection alarm times ranged between 31(0) and 60(22.6) seconds, with no difference between PIP group (H(4) = 9.3, p > 0.05).

The accelerometer displayed pressure-dependent accuracy, increasing with higher PIP, suggesting a minimum pressure threshold is needed for reliable monitoring. App alert times varied greatly, compounding the accuracy of adverse event detection. To determine their true utility, rigorous investigation of neonatal wearable home-monitoring devices is required.