Advancing neonatal brain function monitoring: From sleep and EEG assessments to outcome prediction

Xiaowan Wang

Research output: ThesisDoctoral thesis 1 (Research UU / Graduation UU)

Abstract

Continuous bedside monitoring of brain function is instrumental in implementing timely, brain-oriented care strategies for neonatal patients admitted to the neonatal intensive care unit (NICU), especially for those born prematurely (born at <37 weeks of gestation). However, there are prominent issues that are urgent to be addressed to enable effective brain function monitoring in the NICU. One major challenge is that, despite the important role of sleep in early brain development, there is currently no dedicated method that allows continuous, automated sleep monitoring for preterm neonates in the NICU setting. Another issue concerns the optimal utilization of the electroencephalogram (EEG) in neonatal critical care due to the diversity of EEG parameters available.
In this context, the present thesis aims to advance neonatal brain function monitoring through 1) developing efficient methods for automated sleep monitoring within the NICU environment and identifying clinically applicable outcome predictors based on sleep data (Part I), and 2) identifying neonatal EEG parameters with significant prognostic value and exploring strategies to improve the practical application of neonatal EEG in clinical settings (Part II).
In Part I, We first reviewed the existing literature on the value of cardiorespiratory parameters for sleep classification in preterm infants (Chapter 2). Subsequently, we developed and deployed an automated, real-time bedside sleep staging system based on routinely measured vital signs for preterm infants in our NICU (Chapter 3). Using the developed automated sleep staging algorithm, we further demonstrated the potential of algorithmically derived active sleep as an early predictor for subsequent white matter development in human preterm infants (Chapter 4). Additionally, we demonstrated the feasibility of differentiating sleep–wake states in preterm infants using quantitative analysis of dual-channel EEG, another widely used tool in NICUs (Chapter 5).
In Part II, we first examined the existing literature regarding the associations between EEG parameters and structural brain imaging measurements in premature infants (Chapter 6). Subsequently, we applied machine learning-based regression and classification models to evaluate the predictive ability of quantitative EEG features on multiple long-term neurodevelopmental outcomes by using aEEG in a cohort of extremely preterm neonates (Chapter 7) and multichannel EEG in a cohort of high-risk preterm and term neonates (Chapter 8). Finally, we gathered nurses’ experiences and perspectives regarding the usefulness of EEG monitoring in the NICU and discussed potential avenues for future improvements (Chapter 9).
To conclude, our research presents significant advancements in the domain of neonatal brain function monitoring, with a special focus on preterm infants in the NICU. By addressing gaps in automated sleep monitoring and optimizing the use of EEG parameters, this thesis contributes to the enhancement of brain-oriented care strategies for vulnerable neonatal populations. Furthermore, the application of advanced machine learning techniques in our research opens new avenues for personalized, predictive neonatal care. By integrating these novel approaches into clinical practice, we can improve the early detection of neurological issues, enable more targeted interventions, and ultimately improve future well-being and quality of life for preterm infants.
Original languageEnglish
Awarding Institution
  • University Medical Center (UMC) Utrecht
Supervisors/Advisors
  • Dudink, Jeroen, Supervisor
  • Benders, Manon, Supervisor
  • Tataranno, Maria Luisa, Co-supervisor
Award date7 Oct 2024
Place of PublicationUtrecht
Publisher
Print ISBNs978-94-6506-293-8
DOIs
Publication statusPublished - 7 Oct 2024

Keywords

  • preterm
  • sleep
  • aEEG
  • EEG
  • NICU
  • brain function monitoring
  • brain maturation
  • neurodevelopmental outcome
  • machine learning

Cite this