Neurobehavioral Monitoring: A Multimodal Perspective

ptimal care for newborns in the neonatal intensive care unit (NICU), especially for premature infants, requires innovative methods to monitor vital functions and behavior. Although current monitoring techniques are proven effective, many of these techniques are burdensome, lack personalization, and have limited capabilities. Therefore, these techniques must be further developed and expanded to support more personalized and practical treatment in the NICU. In this context, this dissertation is part of the INFANS project (Integrating Functional Assessment Measures for Neonatal Health), a European collaboration funded by the European Union’s Horizon 2020 program. This research focuses on enhancing real-time clinical decision-making in the early postnatal period through innovative neuro-monitoring in three areas: (1) sleep and vital functions, (2) brain maturation, and (3) personalized circulation management.

The first part of the dissertation investigates the monitoring of sleep and vital signs in newborns and children, integrating techniques such as near-infrared spectroscopy (NIRS), heart rate, respiratory rate, and ultra-wideband radar. By applying machine learning algorithms to the data collected from these techniques, we developed minimally invasive and contactless methods for behavioral (sleep) and vital sign monitoring that are suitable for daily intensive care use.

The second part focuses on assessing brain maturation, which is particularly crucial for newborns at high risk of developmental problems, such as those born extremely prematurely. Multi-channel electroencephalogram (EEG) data collected during the early postnatal days was analyzed using a deep neural network (DNN). This analysis compares predicted with actual age to provide insight into long-term brain development effects.

The final part of the dissertation explores personalized treatment strategies for severely premature infants. This section includes monitoring arterial blood pressure, oxygen saturation, NIRS, and EEG to optimize circulatory treatments and evaluate the risk of brain hemorrhage. These personalized strategies have the potential to improve the long-term outcomes for NICU newborns. In summary, this dissertation examines neuro-monitoring algorithms that combine information from multiple modalities using innovative analysis methods to optimize care for vulnerable newborns.