Determinants of neonatal sepsis outcome: Harnessing machine learning algorithms for enhanced clinical decision-making

Background: Neonatal sepsis is a major cause of morbidity and mortality in low- and middle-income countries where healthcare resources are often limited. Objectives: This study utilizes machine learning (ML) algorithms to identify determinants of neonatal sepsis outcomes and enhance prediction accuracy. Methods: Data on neonatal sepsis cases were gathered from the Neonatal Intensive Care Unit (NICU) at Ekiti State University Teaching Hospital. The target variable was neonatal sepsis outcome, with predictor variables including weight at admission, sex, mode of delivery, place of delivery, number of gestations, and age at admission. The dataset was split into training (80%) and validation (20%) sets. Five ML algorithms—Logistic Regression, Random Forest, Gradient Boosting, Decision Tree, and Extra Trees Classifier—were applied, and their performance was compared based on accuracy, precision, recall, and F1 score. Hyperparameter tuning was conducted using GridSearch. Results: Logistic regression achieved the highest accuracy (0.86), with birth weight, place of delivery, and duration of stay identified as major determinants of neonatal sepsis outcomes. Univariate analysis indicated higher mortality among preterm infants (20.9%) compared to term births (8.1%) and among those born outside hospitals (20.0% vs. 5.6%). Conclusion: Promoting hospital-based deliveries and implementing mandatory neonatal weight assessments could help reduce sepsis-related mortality through early intervention and improved care for at-risk infants.