Millimeter-Wave MIMO radar for contactless neonatal heart-rate assessment: performance under common resuscitation-related maneuvers and feasibility in clinical workflow.

Background: Rapid and reliable heart-rate (HR) assessment is critical during neonatal resuscitation. Contactless monitoring may serve as a valuable adjunct to conventional methods, but the clinical performance of millimeter-wave radar under routine resuscitation-related maneuvers remains unclear. We evaluated the performance and workflow feasibility of a millimeter-wave multiple-input multiple-output (MIMO) radar system for contactless neonatal HR assessment. Methods: In this single-center prospective study, we conducted two sequential experiments. In the factor-quantification experiment, 60 clinically stable neonates in the neonatal intensive care unit (NICU) were monitored simultaneously with radar and a 3-lead ECG bedside monitor as the reference. We assessed HR measurement performance under quiet baseline conditions and four simulated resuscitation-related maneuvers: drying, oronasal suctioning, auscultation, and postural change. Measurement error was quantified using root-mean-square error (RMSE), and agreement was evaluated using Bland-Altman analysis. In the real-workflow experiment, 26 neonates on a radiant warmer in the operating room were enrolled. We analyzed 115 auscultation windows by comparing mean radar-derived HR with auscultated HR at the window level and by stratifying per-second radar-derived HR errors within each window. Results: Under quiet baseline conditions, radar-derived HR showed agreement with ECG, with most RMSE values below the predefined 10-bpm descriptive benchmark. Measurement performance was relatively preserved during simulated drying, suctioning, and auscultation. In contrast, postural change significantly increased error and caused systematic HR underestimation. In the real-workflow experiment, window-level differences between mean radar-derived HR and auscultated HR within auscultation windows were centered near 0 bpm. In the per-second error analysis within auscultation windows, errors of ≤5 bpm were the most frequent category, whereas errors of >15 bpm were uncommon. Conclusion: Millimeter-wave MIMO radar showed relatively small HR measurement error for contactless short-window assessment under quiet baseline conditions, particularly when the infant remained stable and oriented toward the sensor. Postural change was the dominant source of increased measurement error and may limit reliability during thoracic reorientation. These findings support the potential role of radar as an adjunct tool for short-window HR assessment during neonatal care and resuscitation-related conditions. Continuous beat-to-beat accuracy was not validated in this study and requires further investigation using continuous ECG-based reference standards.