Artificial intelligence techniques in blood banks — a systematic review of predictive innovations

Cardona D.C.V.; Cardona-Acevedo S.; Valencia-Arias A.; Valencia J. · Acta Haematologica Polonica (2025) · DOI: 10.5603/ahp.104427

Resumen

In the field of healthcare, blood banks play a pivotal role in overseeing the collection, storage, and distribution of blood in critical situations. However, they encounter substantial challenges in the management of resources, including demand forecasting and inventory optimization. In addressing these issues, artificial intelligence (AI) is emerging as a promising solution. This study analyzes the distribution and impact of AI methodologies in blood banks, identifying regression techniques, support vector machines, and transformer-based models as the most frequently applied approaches. Key areas of improvement include blood supply forecasting, resource allocation, and decision accuracy, which are the most prevalent impacts observed in demand prediction. Additionally, AI contributes to transfusion safety by enhancing transfusion decision accuracy and blood supply reliability, while predictive models for genetic blood typing and hemoglobin-based matching optimize donor-recipient compatibility. Despite these advances, significant challenges persist, particularly regarding data quality, model accuracy adaptation, and AI explicability, underscoring the necessity for robust technological infrastructure and specialized personnel training. Addressing these barriers through clear policies and a structured integration strategy is essential to fully harness AI’s potential in transfusion medicine. Copyright

Instituciones: Universidad Ricardo Palma