Blood Transfusion Database for Pediatric Congenital Heart Disease Surgery

This dataset contains perioperative data from 3,342 pediatric patients undergoing congenital heart disease surgery. Variables include demographics (age, gender, weight), preoperative laboratory values (hemoglobin, hematocrit, platelet count, coagulation parameters, liver function tests), surgical complexity (Aristotle Score), cardiopulmonary bypass use, and intraoperative blood transfusion volumes (RBC, plasma, platelets). The transfusion rates were 17.7% for RBC, 31.3% for plasma, and 2.0% for platelets. Missing values are denoted by "/". This data supports the development of machine learning models for predicting blood transfusion requirements in pediatric cardiac surgery. Missing data primarily occurred in coagulation parameters (PT, APTT) and some laboratory values that were not routinely measured in all patients. This study was approved by the Institutional Review Board. The requirement for informed consent was waived due to the retrospective nature of the study. All data were de-identified to protect patient privacy.
This dataset is associated with the manuscript: "Machine Learning-Based Prediction of Blood Transfusion in Pediatric Congenital Heart Surgery: A Novel Framework for Unified Comparison"