Predictive Analysis of Healthcare-Associated Blood Stream Infections in the Neonatal Intensive Care Unit Using...

Area of Science:

• Neonatal Medicine
• Infectious Diseases
• Medical Informatics

Background:

• Healthcare-associated infections (HAIs) pose significant risks to neonates, with increasing mortality linked to preterm births and childbirth complications.
• Despite advancements, identifying factors influencing neonatal mortality and morbidity remains challenging.
• This study focused on HAIs within a level III Neonatal Intensive Care Unit (NICU).

Purpose of the Study:

• To identify risk factors associated with healthcare-associated bloodstream infections (HABSIs) in neonates.
• To develop predictive models for HABSI to aid in prevention and management.
• To analyze the impact of HABSIs on neonatal outcomes.

Main Methods:

• Statistical analyses and logistic regression were employed to determine associations between risk factors and HABSIs.
• A supervised machine learning approach was used to predict HABSI occurrence.
• Seven artificial intelligence models were evaluated for predictive performance.

Main Results:

• Birthweight and the duration of central line catheterization were identified as the most significant predictors of HABSI.
• Neonates with HABSIs exhibited lower gestational age and birthweight.
• HABSIs correlated with extended hospitalization and longer durations of umbilical and central line catheterization.

Conclusions:

• Birthweight and central line catheterization duration are significant predictors of HABSI in neonates.
• Logistic Regression (LR) and Multi-layer Perceptron (MLP) models demonstrated superior performance in predicting HABSIs, effectively handling imbalanced data.
• These findings underscore the importance of monitoring birthweight and catheter use to mitigate HABSI risks.