Prenatal predictors of neonatal intensive care unit admission due to respiratory distress

Machine Learning to Improve Accuracy of Transcutaneous Bilirubinometry

INTRODUCTION

This study aimed to develop models for predicting total serum bilirubin by correcting errors of transcutaneous bilirubin using machine learning based on neonatal biomarkers that could affect spectrophotometric measurements of tissue bilirubin.

METHODS

This retrospective study included infants born at our hospital (≥36 weeks old, ≥2,000 g) between January 2020 and December 2022. Infants without a phototherapy history were included. Robust linear regression, gradient boosting tree, and neural networks were used for machine learning models. A neural network, inspired by the structure of the human brain, was designed comprising three layers: input, intermediate, and output.

RESULTS

Totally, 683 infants were included. The mean (minimum-maximum) gestational age, birth weight, participant age, total serum bilirubin, and transcutaneous bilirubin were 39.0 (36.0-42.0) weeks, 3,004 (2,004-4,484) g, 2.8 (1-6) days of age, 8.50 (2.67-18.12) mg/dL, and 7.8 (1.1-18.1) mg/dL, respectively. The neural network model had a root mean square error of 1.03 mg/dL and a mean absolute error of 0.80 mg/dL in cross-validation data. These values were 0.37 mg/dL and 0.28 mg/dL, smaller compared to transcutaneous bilirubin, respectively. The 95% limit of agreement between the neural network estimation and total serum bilirubin was -2.01 to 2.01 mg/dL. Unnecessary blood draws could be reduced by up to 78%.

CONCLUSION

Using machine learning with transcutaneous bilirubin, total serum bilirubin estimation error was reduced by 25%. This integration could increase accuracy, lessen infant discomfort, and simplify procedures, offering a smart alternative to blood draws by accurately estimating phototherapy thresholds.

Maternal anemia and long-term respiratory morbidity of the offspring-Results of a population-based cohort

PURPOSE

To evaluate the association between prenatal maternal anemia (hemoglobin, Hb < 11 g/dl) and long-term respiratory morbidity of the offspring.

METHODS

A retrospective population-based cohort analysis was performed at a single tertiary medical center. We include all singletons born between 1991-2014 and discharged alive. Offspring with congenital malformations were excluded. The 3 study groups were defined on maternal Hb levels on discharge: severe anemia (< 7.0 g/dl), mild to moderate anemia (7.0-11.0 g/dl) and unexposed (≥ 11.0 g/dl). Offspring respiratory morbidity was predefined on ICD-9 codes and recruited from the hospitalized medical records. A Kaplan-Meier survival curve was formed to compare the cumulative hospitalization and a multivariable Cox survival analysis was used to control for cofounders (gestational age, maternal age, diabetes, hypertensive disorders, post-partum hemorrhage and transfusion of blood products).

RESULTS

214,305 deliveries met the inclusion criteria: 807 (0.3%) mothers had severe anemia, 105,196 (49.1%) mothers had mild-moderate anemia, and the remaining were not anemic (108,302, 50.5%). Respiratory hospitalization was significantly higher among the offspring born to anemic mothers (6.2%; 5.3% and 5.1%; p = 0.020, in the study groups). However, association between maternal anemia and respiratory-related hospitalization remained significant only among the mild-moderate anemic group (adjusted Cox hazard ratio = 1.1; 95% CI 1.05-1.14; p < 0.01). The Kaplan-Meier survival analysis showed significant higher total respiratory hospitalizations in offspring according to maternal anemia status (Log Rank p-value =  < .001).

CONCLUSION

Maternal anemia was associated with long-term respiratory morbidity of the offspring.

Impact of Technology on Neonatal Intensive Care Unit Admissions and Length of Stay: A Retrospective Study

Background Neonatal intensive care units (NICU) provide essential medical care to neonates; however, they are associated with hospital-acquired infections, less maternal-newborn bonding, and high costs. Implementing strategies to lower NICU admission rates and shorten NICU length of stay (LOS) is essential. This study uses causal-inference methods to evaluate the impact of care managers using new technology to identify and risk stratify pregnancies on NICU admissions and NICU LOS. The NICU LOS will decrease as a result of the use of new technology by care managers. Study design This retrospective study utilized delivery claims data of pregnant women from the CareFirst BlueCross BlueShield Community Health Plan District of Columbia from 2013 to 2022, which includes the pre-intervention period before the use of new technology by care managers and the post-intervention period with the use of new technology by care managers. Our sample had 4,917 deliveries whose maternal comorbidities were matched with their neonate's outcomes. Methods To evaluate the impact of the technological intervention, both Generalized Linear Models (GLMs) and Bayesian Structural Time-Series (BSTS) models were used. Results Our findings from the GLM models suggest an overall average reduction in the odds of NICU admissions of 29.2% and an average decrease in NICU LOS from 7.5%-58.5%. Using BSTS models, we estimate counterfactuals for NICU admissions and NICU LOS, which suggest an average reduction in 48 NICU admissions and 528 NICU days per year. Conclusion Equipping care managers with better technological tools can lead to significant improvements in neonatal health outcomes as indicated by a reduction in NICU admissions and NICU LOS.