Abstract Introduction In the neonatal intensive care unit (NICU), transthoracic impedance is used to monitor breathing and detect apneas in preterm infants. However, this method does not measure airflow and existing airflow-monitoring tools are unsuitable for continuous bedside use. To address this, we developed a small wireless broadband acousto-mechanical sensor (BAMS) integrating a microphone to capture breath sounds and an inertial measurement unit (IMU) to capture breathing efforts. This study aimed to evaluate (1) the BAMS microphone for detecting respiratory airflow compared with a thermistor, and (2) the BAMS IMU for detecting breathing efforts compared with respiratory inductance plethysmography (RIP) in preterm infants. Methods This is Phase 2 of a prospective observational pilot study (ClinicalTrials.gov—NCT05196646). Twenty spontaneously-breathing preterm infants (gestational age 32weeks; corrected age 37weeks) were recruited from the Montreal Children’s Hospital NICU. A BAMS placed on the suprasternal notch recorded respiratory sounds and breathing effort (index signals) for 3h. A nasal thermistor and RIP captured gold standard measures of respiratory airflow and breathing effort (reference signals). The power between 150-450Hz in the raw acoustic signal was computed over 0.5sec-windows to derive a respiratory airflow waveform. An algorithm identified noninterpretable segments (due to movement artefact, technical limitations) for exclusion prior to analysis. Respiratory rates (RRs) were estimated using peak-detection methods over 60-second windows and compared between index and reference signals using Bland-Altman and Clarke error grid analyses. Results Participants had a median IQR gestational age of 28.4 27.1, 29.7 weeks, corrected age of 35.3 33.6, 35.6 weeks, and study weight of 2165 1883, 2533 grams. Due to missing data, 19 infants were included for microphone analyses and 18 for IMU analyses. The total interpretable data-time in minutes (% of recorded data) for the thermistor, microphone, RIP, and IMU was 2878 (81.9%), 2453 (82.7%), 2746 (78.1%), and 2151 (77.1%), respectively. The mean difference between thermistor- and microphone-derived RRs was 0.5 bpm (Limits of Agreement, LOA: -8.0, 9.0; n = 2190), and between RIP- and IMU-derived RRs was 1.2 bpm (LOA: -5.6, 7.9; n = 1955).Using Clarke error grid analyses, 97.4% and 98.6% of datapoints from the microphone and IMU fell within clinically acceptable differences from their respective reference values (±20%; Fig.1). Conclusion The BAMS showed high agreement with reference measures of respiratory airflow and breathing efforts in stable preterm infants. This wireless wearable technology holds the potential of enabling the characterization of apneas in the NICU while minimizing the inconvenience associated with wired technology. This abstract is funded by: the Pediatric Research Foundation and the Montreal Children’s Hospital Foundation
Jeanne et al. (2026) studied this question.