Introduction
Approximately 9 in 1000 children are born with congenital heart disease (CHD), of whom a quarter are classified as critical CHD (CCHD) and require an intervention within their first year. Monitoring these patients in the Paediatric Intensive Care Unit (PICU) is crucial, yet with increasing amounts of data, detecting subtle changes that are important for the disease progression interpretation of all vital signs becomes difficult, even for skilled physicians. Machine learning (ML) offers potential solutions, however, challenges such as inter-patient variability and the absence of clear definitions for haemodynamic instability persist. This study aims to develop a ML algorithm for early prediction of haemodynamic instability in CCHD patients with high frequency vital-signs, addressing these challenges through objective labelling methods and stratification approaches.

Methods
Two approaches, on population and patient level, were developed with nested cross-validation (CV). Due to a high inter-patient variability, the patient specific approach was added. A first iteration of objectively labelling haemodynamic instability was proposed, based on medical interventions such as medication administration and fluid therapy. Since it is difficult to retrospectively determine for how long patients were unstable, multiple values for instability duration (dT ) were added to the analysis.
To capture the temporal dependency of time-series data, lag-analysis was performed, adding the relation between the vital signs and their previous values to the model development. Lag-analysis included a sliding window that moved over the data. The width of sliding window (W ) was optimised during the model development. Additionally, a horizon (r ) was implemented, so the data within the sliding window were predicting future timestamps.

Results
This retrospective study included a total of 224 admissions in the analysis. Two random forest classifiers were trained using a nested CV structure to detect haemodynamic instability in CCHD patients. For both approaches the same temporal settings (W : 50 minutes, r : 45 minutes, dT: 120 minutes) were used. This study has shown that the between-patient approach had notable differences between the mean train (85%, AUCPR) and test performance (40%, AUCPR). The in-patient approach, while using 20% and 10% of the test data for training, still yielded a test performance of 96% (AUCPR) and 90% (AUCPR), respectively.

Discussion and conclusion
Generally speaking, the experiments suggest that the first iterations of the models were not robust and generalised poorly. It is most likely caused by a large inter-patient variability and a simple labelling system that is still depending on subjectivity.
This study has shown that the proposed prediction model, which combines high frequency vital signs, labels, and temporal settings (W, r, dT ), requires additional refinement before it can be considered clinically feasible to implement this model as a reliable bedside tool for predicting haemodynamic instability.

Introduction: In the Pediatric Intensive Care Unit (PICU), maintaining stable cerebral blood flow (CBF) is essential incirtically ill children, as fluctuations can lead to severe neurological complications. Cerebral autoregulation (CA) ensures stable CBF despite changes in systemic blood pressure. Traditional methods such as intracranial pressure (ICP) monitoring and transcranial Doppler (TCD) provide indirect assessments of CA and are often invasive or limited in continuous monitoring. Near-infrared spectroscopy (NIRS) offers a more non-invasive alternative by continuously measuring cerebral oxygenation. By correlating cerebral regional oxygen saturation (CrSO2) with, though invasive, mean arterial pressure (MAP), the cerebral oximetry index (COx) can be calculated, quantifying the brain’s autoregulatory efficiency. This allows clinicians to determine the optimal MAP (MAPopt) per individual, where CA is most effective, which could improve patient outcomes by maintaining stable cerebral perfusion.
Objective: This thesis aims to evaluate COx, derived from (CrSO2) and MAP, to identify the MAPopt in pediatric patients undergoing surgery for congenital diaphragmatic hernia (CDH) or esophageal atresia (EA). Additionally, it examines differences in CA across pre-, intra-, and postoperative phases, providing insights into how surgery affects CA.
Method: Left and right CrSO2 and MAP signals, pre-, intra-, and post-operatively, were retrospectively analyzed to assess CA using the COx. Data preprocessing involved the removal of artifacts, normalization, and signal alignment. COx was calculated as the moving Pearson correlation of CrSO2 with MAP over time
using a 5 minute sliding window with a one minute stepsize. Impaired CA was defined as a COx cut-off value ≥ 0.3, defining a correlation between the MAP and CrSO2. The MAPopt was identified as the MAP associated with the lowest COx values, representing the most effective CA. The optimal MAP range was calculated per patient.
Results: CA was interpreted to be most compromised during the intraoperative phase, with elevated COx values in both CDH and EA patient groups. The MAPopt was successfully determined in most cases, showing variability across operative states, with lower values observed intraoperatively. The percentage of time with COx above the cut-off of 0.3 was highest during surgery, especially for EA patients. No remarkable differences were found between the left and right CrSO2 signals.
Conclusion: The study demonstrates the feasibility of using NIRS-derived COx to determine MAPopt in neonates undergoing CDH or EA surgery. These findings highlight the potential for individualized MAP targets
to optimize cerebral perfusion, especially during critical surgical interventions.