ABMES

Background: Premature infants are at high risk for medical adhesive-related skin injuries. It is crucial to establish reliable methods for non-contact monitoring. This study aimed to identify the optimal piezoelectric sensor placement (chest, abdomen, and hips) for respiratory rate estimation in each body position (supine, lateral, and prone) and demonstrate that body movements affect respiratory estimation accuracy.

Methods: Respiratory rates were estimated for 30 infants admitted to the Growing Care Unit for each body position. Body positions, movements (categorized by the Neonatal Behavioral Assessment Scale), and sensor placements were recorded and analyzed for their effects on accuracy. Estimation accuracy was defined as the percentage of time during which the error between the estimated and reference (respiratory rate displayed on the monitor) was within 5 breaths/min.

Results: The most improved estimation accuracies were achieved in the supine position with the chest sensor, in the lateral position with the abdominal sensor. The prone position was limited to a single case, thus rendering it inappropriate to draw comparisons. Accuracy declined progressively as the infants’ state changed from sleeping to crying. Moreover, substantial reductions in accuracy were observed during full-body movements (quiet and active alert state) as well as during arm movements (drowsy and quiet alert state).

Conclusion: This study demonstrated that the accuracy of non-contact respiratory rate estimation using piezoelectric sensors is highly impacted by body position, movement, and sensor placement. Optimizing the algorithm to systematically account for these variables is essential for improving the robustness and accuracy of non-contact vital sign monitoring.