Related Experiment Video
Updated: Jul 12, 2025
06:19
Author Spotlight: Implications of Non-Nutritive Sucking on Speech Emergence and Infant Development
Published on: April 19, 2024
796
Spectral features of non-nutritive suck dynamics in extremely preterm infants
Steven M Barlow1, Chunxiao Liao2, Jaehoon Lee3
1Department of Communication Disorders and Department of Biological Systems Engineering, Center for Brain, Biology & Behavior, University of Nebraska, Lincoln, NE, USA.
Pediatric Medicine (Hong Kong, China)
|October 30, 2023
Summary
This study modeled non-nutritive suck (NNS) patterns in preterm infants using frequency domain analysis and machine learning. Findings reveal distinct NNS developmental trajectories, offering insights into neural regulation and feeding readiness in the neonatal intensive care unit (NICU).
Area of Science:
- Neonatal Physiology
- Neurodevelopmental Pediatrics
- Computational Neuroscience
Background:
- Non-nutritive suck (NNS) is crucial for oral feeding readiness in preterm infants within the neonatal intensive care unit (NICU).
- Current time-domain NNS measures offer limited insight into the underlying suck pattern generation.
- Frequency-domain analysis using Fourier and machine learning (ML) techniques remains underexplored for NNS development in extremely preterm infants (EPIs).
Purpose of the Study:
- To model the developmental changes in non-nutritive suck (NNS) patterns in extremely preterm infants (EPIs).
- To apply Fourier transforms and machine learning (ML) techniques for frequency-domain analysis of NNS.
- To identify distinct NNS developmental trajectories in EPIs.
Main Methods:
- 117 extremely preterm infants (EPIs) received pulsed or sham orocutaneous intervention during tube feedings for 4 weeks.
- Non-nutritive suck (NNS) signals were digitized and analyzed in the frequency domain using Welch and Yule-Walker power spectral density (PSD) methods.
- Machine learning (ML) cluster analysis identified distinct NNS growth patterns, followed by linear mixed modeling (LMM) to assess influencing factors.
Main Results:
- Machine learning (ML) analysis revealed three distinct classes of non-nutritive suck (NNS) development in the frequency domain.
- Key NNS frequency-domain measures (peak frequency, PSD amplitude, AUC) were significantly influenced by postmenstrual age (PMA).
- Respiratory status (RDS, BPD) and intervention type did not significantly impact these frequency-domain NNS characteristics.
Conclusions:
- Frequency-domain analysis, particularly using ML, effectively characterizes distinct non-nutritive suck (NNS) developmental patterns in preterm infants.
- Understanding NNS rhythmogenesis in the frequency domain provides deeper insights into the suck central pattern generator (sCPG) in EPIs.
- Identifying invariant versus modifiable NNS features can inform improved therapeutic strategies for feeding interventions in preterm infants.
