A Decision-Tree Approach to Assist in Forecasting the Outcomes of the Neonatal Brain Injury

Area of Science:

• Neonatal neurology
• Pediatric critical care
• Medical informatics

Background:

• Neonatal encephalopathy (NE) is a major cause of mortality and morbidity in newborns, leading to severe neurological deficits.
• Existing scoring systems for NE prognosis have limitations, necessitating complementary predictive tools.
• Clinical risk factors like birth weight and Apgar scores are known predictors, but their integrated predictive power needs refinement.

Purpose of the Study:

• To propose and evaluate a decision-tree approach for predicting abnormal neurological outcomes in newborns with NE.
• To utilize clinical risk factors for a more accessible and complementary prognostic tool.
• To enhance the early identification of high-risk neonates requiring timely intervention.

Main Methods:

• Retrospective study of 188 newborns with perinatal encephalopathy and seizures.
• Application of standard statistical methods (mean, median, odds ratios) for risk factor analysis.
• Implementation of Classification and Regression Trees (CART) and cluster analysis for outcome prediction.

Main Results:

• 84 out of 188 neonates experienced abnormal outcomes.
• Cerebrovascular impairments, metabolic anomalies, and infections were the leading etiologies.
• CART analysis categorized neonates into high-risk (75-100%), intermediate-risk (52.9%), and low-risk (0-25%) groups, differentiating preterm and full-term outcomes.

Conclusions:

• Decision-tree approaches offer a valuable first-step tool for neonatal encephalopathy prognosis.
• This method complements existing scoring systems by providing a clear risk stratification based on clinical factors.
• The findings highlight the potential for improved early risk assessment and management of NE.