Exploring the Characteristics of Infants That Influence Their Number of Transfusions Using a Multivariable...

Area of Science:

• Neonatal care
• Medical informatics
• Predictive modeling

Background:

• Neonatal transfusions are vital but influenced by complex perinatal factors.
• Understanding these factors is key for timely infant care and management.

Purpose of the Study:

• To develop predictive models for infant transfusions using perinatal data.
• To identify key factors influencing the need for neonatal transfusions.

Main Methods:

• Collected data from electronic medical records of infants.
• Categorized infants based on the number of transfusions received.
• Constructed and evaluated multivariable, random forest, and gradient boosting tree models to predict transfusion needs.

Main Results:

• Underweight status, premature birth, Apgar scores, and gestational diabetes mellitus significantly impacted transfusion requirements.
• Random forest models showed strong predictive performance (AUC up to 0.8037 on training sets).
• Apgar scores at 1, 5, and 10 minutes demonstrated potential as preliminary predictive factors.

Conclusions:

• Apgar scores can inform early transfusion strategies for newborns.
• Further model optimization is needed for accurate blood transfusion prediction in infants due to data limitations.