Background: Neurocritical care in the neonatal intensive care unit is centred on improving neonatal and paediatric outcome by the prevention of new or worsening brain injury in preterm and term newborn infants. Cerebral near-infrared spectroscopy is a non-invasive tool for the continuous monitoring of cerebral oxygen saturation (rcSO2) that could be exploited as an index of brain health to assist neonatologists in predicting and diagnosing brain injury, helping to inform therapeutic strategies. Objective: Universal threshold-based approaches to rcSO2 to infer hypoxia may be flawed as they do not account for inter-patient variability, which is significant in the NICU infant. In contrast, we sought to apply machine learning methods to NIRS data.
Methods & results: We have analysed NIRS data from infants using machine learning. XGBoost, also known as Extreme Gradient Boosting, is a supervised learning technique that uses an ensemble approach based on the Gradient boosting algorithm. We have shown that cerebral NIRS features were associated with MRI abnormalities and/or death (AUC: 0.722, 95% CI: 0.579 - 0.849). Furthermore, we are currently exploring the application of supervised deep learning algorithms to cerebral NIRS data. Our convolutional neuronal networks (CNN) model consists of multiple convolutional layers, followed by a fully connected (FC) layer. We report on potential challenges in applying deep learning approaches to NIRS data.
Conclusion: Applying supervised deep learning algorithms to cerebral NIRS data offers potential benefits in the assessment of brain health, as it seeks to identify data-driven features and can learn iteratively from new data.

Background
Tracheal intubation (TI) is a core skill for Neonatal-Perinatal Medicine (NPM) fellows to develop. NPM fellows attain TI competency at variable times during fellowship. The best training structure to support TI competency development is unknown.
Objective
1. Characterize US NPM fellowship TI education & assessment practices.
2. Explore how NPM fellowship size & educational resources impact TI education.
Methods
We conducted a cross-sectional study of NPM fellowship leaders in Sept 2022. A 32-item survey was created using iterative review by 6 NPM education experts. Survey questions aimed to understand TI education practices, barriers, assessment practices, and perceived effectiveness. The survey was sent via REDCap.
Results
89% (98/110) of fellowship programs responded to the survey (Table 1). Large programs report higher airway curriculum satisfaction (p=0.009) and more hands-on TI training during clinical orientation (p=0.012) than small and medium programs. Programs with highest curriculum satisfaction report fewer time-constraints (p=0.012) and equipment shortages (p=0.036) (Table 2). There is wide variability in preferred definitions for TI competency among all programs (Figure 1). TI competency is typically reviewed every 6 months. The number of TI attempts (median=50, IQR 36.25 to 63.75) for graduating fellows was variable.
Conclusion
Significant variability exists in educational curricula and resources for TI training in NPM programs. Large programs report greater curricular satisfaction. Trainee TI competency is poorly defined & reviewed infrequently. Programs report variable numbers of TI completion by graduation. Future investigation is needed to determine which TI educational & assessment practices correlate with improved clinical performance.

Introduction:
We present the findings of a retrospective audit looking into the management of prolonged neonatal jaundice cases.
Prolonged neonatal jaundice is a common cause of referral to paediatric OPD. It is defined as jaundice persistent beyond 14 days in the term infant or 21 days in the preterm infant.
It is commonly encountered in breastfeeding infants.

Methods:
A survey monkey was developed and the audit department identified a total of 81 cases referred in the preceding 12 months.
Data collection included demographics, type of feeding, gestation, weight, stools colour, urine colour, previous requirement for phototherapy, investigations and tests' results.

Results:
68 cases were term infants and 13 were born prematurely. 88% of cases were breastfed infants (Chart 1).
3% of cases did not have bilirubin levels measured. 92% of cases had conjugated bilirubin levels < 20 micromol/L.
In 5% of cases, conjugated bilirubin levels were greater than 20 micromol/L (Chart 2). 4 out 81 cases had raised conjugated bilirubin.


Discussion:
In our study, 91% of the cases were breastfed and breast milk jaundice was the final diagnosis. Biliary atresia was identified in 2 cases. CMV infection and TPN were the potential cause in one case and one case was still under investigation to identify the underlying cause.
Awareness of national, regional and local guidelines for management of prolonged neonatal jaundice is crucial to avoid delay in diagnosing treatable causes e.g. hypothyroidism. Early diagnosis and timely referrals are crucial in cases with biliary atresia to avoid bad outcomes.

Background
Neonatal probiotic supplementation remains contentious. Benefit may be restricted to supplementation with particular bacterial strains. We present data from infants born at Cork University Maternity Hospital, Ireland, supplemented with Bifidobacterium bifidum and Lactobacillus acidophilus (Infloran®).

Objective
To evaluate the combined outcome of death or severe necrotising enterocolitis (NEC) before and after introduction of routine probiotic supplementation to very preterm infants.

Methods
A retrospective study of infants <32 weeks gestation and <1500g surviving beyond 72 hours of life was performed. Two 6-year epochs; pre-probiotics (Epoch 1: 2008-2013) and with probiotics (Epoch 2: 2015-2020), were evaluated. The primary outcome was defined as death after 72 hours or NEC Bell stage 2a or greater.

Results
Seven-hundred-and-forty-four infants were included (Epoch 1: 391, Epoch 2: 353). The primary outcome occurred in 67 infants (Epoch 1: 37, Epoch 2: 30, p=.646)). After adjustment, the difference was significant ((OR(95% CI): 0.53 (0.29 to 0.97), p=.038). Differences between epochs did not depend on gestational age group (< 28 weeks; ≥ 28 weeks).

Discussion
Routine administration of a B. bifidum and L. acidophilus probiotic at our institution was associated with reduction in severe grade NEC and/or Death after adjustment for confounding variables.

Background: Maternal thrombocytopenia during pregnancy is common. The relationship between maternal and neonatal thrombocytopenia is poorly understood.

Objectives: To determine whether a correlation exists between neonatal and maternal platelet counts (<100 X 10^9/L) given the non-evidence-based practice of screening these neonates with a full blood count (FBC).

Methods: We identified records from 557 thrombocytopenic mothers and the 337 associated newborn charts from 2018 to 2022.
Mothers <100 X 10^9/L prior to delivery during present gestation were included. Any thrombocytopenia that occurred outside of pregnancy or in the post-partum period was excluded.

Receiver operator characteristic (ROC) curves were generated using the ‘pROC’ statistical package on R and Area under the Curve (AUC) examined. Coordinates of the “best” fit model were examined using the ‘pROC’ statistical package.

Results:A total of 550 FBCs were taken in neonates of mothers with thrombocytopenia. 16 neonates with clinically significant thrombocytopenia (<100X10^9/L) were identified. “Number(s) needed to treat” (NNT) was 78 FBCs for infants requiring treatment and 183 FBCs to identify a clinically significant hemorrhage.

Maternal platelet count alone was not a significant predictor of neonatal thrombocytopenia (figure 1). The addition of trimester of onset of thrombocytopenia did not significantly improve predictive power. The coordinates of the best platelet count threshold for this dataset were then derived from the ROC curve and determined that a threshold of 78 x10^9/L maternal platelets offered the best sensitivity (74%) and specificity (68%) in this cohort.

Conclusion:Screening FBCs based on maternal platelet counts <100X10^9/L has a poor diagnostic yield and accuracy.

Background: Provision of kangaroo mother care (KMC) in neonatal intensive care units (NICU) is an integral part of neonatal care.

Objective: This study aimed to assess whether there was an optimal position (30 versus 60 degrees) for performing KMC.

Methods: Single centre cross-over randomised controlled trial in a tertiary neonatal intensive care unit. Infants with a minimum corrected gestational age of 28 weeks and minimum 600 grams were included. Participants were randomly assigned to commence KMC at either a 30 or 60 degree angle. The primary outcome measure was the difference in median cerebral near-infrared spectroscopy (NIRS) values between the two angles. The secondary outcomes were median peripheral saturations and median heart rates. The results were analysed using the non-parametric Wilcoxon signed rank test comparing 30 degree and 60 degree angle positions.

Results: Twenty patients were included in the final analyses: median gestational age (GA) was 28+1 weeks (range: 23+2 to 32+6 weeks) and median birth weight was 0.985kg (range: 0.620kg to 2kg). There were no statistically significant differences (p = 0.810) between the median NIRS values at 30 degrees (median rSO2 = 67.5, IQR = 58.3 – 73.8) and 60 degrees (median rSO2 = 68, IQR = 60.5 – 76). There were no statistically significant difference in the median peripheral saturations (p = 1), or median heart rates (p = 0.662) between the two angles.

Conclusion: Maternal positioning at a 30 or 60 degree incline does not impact on cerebral oxygenation values in very preterm infants.