Accurate prediction of blood transfusion requirements is crucial for patient outcomes and resource management in clinical settings. We developed a machine learning model using XGBoost to predict the need for a blood transfusion 2 hours in advance based on up to 7 hours of prior data from two large databases, MIMIC-IV and eICU-CRD. Our federated model showed promising results, with F1 scores of 0.72 and 0.66, respectively.