Sedation monitoring in premature infants

Theoretical framework Several clinical and environmental factors could alter brain development in premature infants. These patients, in fact, are exposed to a consistent number of procedures during their entire time of hospitalization. The use of analgesic and sedative drugs is essential in order to grant them a maximal level of comfort. However, the administration of such drugs is complicated by the level of physiological maturity and by the fact that this special collective of patients is still in a preverbal stage of development. Objective methods like a conventional EEG could help understand more about an infants level of sedation. However, its interpretation is time- consuming and requires a given level of expertise. Today, new methods could be used to automatically detect important EEG features. Machine learning, in fact, gives us the opportunity to recognize important EEG trends that could be further easily interpreted by the care-taking team. Hypotheses The overarching aim of this study is to use deep-machine-learning algorithms for the interpretation of suddenly changes in the EEG-background activity related to the administration of sedation; and to contextualize automatic EEG-background changes related to sedation with the clinical opinion of sedation expressed by the scoring of the Neonatal, Pain, Agitation and Sedation Scale (N-PASS). Methods In this Study, 50 preterm infants undergoing clinical procedures for which sedation is required for a short period of time (e.g. central venous catheter), will be prospectively recruited. Both the deviation from actual gestational age and IBI duration will be used to understand changes in the EEG-background activities during sedation administration. In more, trend-EEG parameters will be correlated to the clinical expert opinion of the level of sedation measured through the N- PASS. Level of originality The topic addressed in this project is of outstanding relevance in neonatology as the use of automatic EEG-trends could be useful for the automatic identification of critical neurological events, to evaluate and ameliorate sedation administration in critically ill infants.

Understanding sedation in newborn infants through brain activity: This project showed that brain activity can provide valuable additional information about how newborn infants respond to pain-relieving and sedative medications. By combining brain monitoring with modern computer-based analysis, we developed new approaches that may in the future help clinicians assess sedation more objectively and individually. Newborn infants receiving intensive care frequently undergo necessary diagnostic and therapeutic procedures. Pain-relieving and sedative medications are therefore an essential part of neonatal care. However, determining how strongly an individual infant is affected by these medications remains challenging. Clinicians currently rely mainly on behavioural signs such as movements, facial expressions and reactions to stimulation. These observations are important, but they provide only indirect information about what is happening in the brain. In this project, we recorded the electrical activity of the brain using electroencephalography (EEG), a non-invasive technique routinely used in neonatal intensive care. We studied how this activity changes following clinically required pain-relieving or sedative medication. Importantly, we compared these changes with each infant's own baseline brain activitiy. This individual baseline proved particularly informative because brain activity naturally differs considerably between newborn infants and changes rapidly with brain maturation. Using automated analysis and machine-learning methods, we were able to identify patterns of brain activity associated with medication administration and with clinically assessed levels of sedation. The project also contributed to the development and refinement of automated measures describing important characteristics of the newborn brain, including its maturation and patterns of activity between individual brain bursts. The resulting prospective dataset covers infants at different stages of brain maturation, allowing these methods to be investigated across a broad developmental range. These findings provide a foundation for a future generation of neonatal brain-monitoring tools. Such methods are not intended to replace clinical observation. Instead, combining clinical assessment with objective information derived directly from brain activity could ultimately help clinicians tailor pain relief and sedation more precisely to the individual infant, while avoiding both insufficient treatment and unnecessary sedation.