Artificial Intelligence Implementation in Transfusion Medicine: Addressing the Challenges of Clinical Adoption

Area of Science:

• Biomedical Informatics
• Health Services Research
• Clinical Pathology

Background:

• Artificial intelligence (AI) and machine learning (ML) are increasingly recognized for their potential to improve transfusion and patient blood management.
• Despite promotion, real-world integration of AI/ML in these clinical areas remains limited.

Purpose of the Study:

• To review exemplar studies on AI/ML deployment in transfusion and patient blood management.
• To identify translational features, barriers, and enablers for AI/ML integration into clinical workflows.

Main Methods:

• A systematic search of PubMed and Web of Science for articles from January 2022 onward was conducted.
• Three prospective studies with workflow integration were selected for detailed analysis of AI/ML implementation.

Main Results:

• Exemplars included a lab tool for ferritin prediction, a patient app for hemoglobin estimation, and a clinician tool for trauma resuscitation needs.
• Common enablers were clinical need alignment, existing infrastructure, interpretable models, and stakeholder engagement.
• Key barriers included data quality, generalizability, and lack of economic evaluation; no study showed improved clinical outcomes or cost.

Conclusions:

• Successful AI/ML adoption requires seamless integration into routine workflows with robust safety, monitoring, and regulatory plans.
• Future research should prioritize implementation frameworks, evaluate downstream impacts, and focus on scalable solutions like lab-embedded analytics and digital patient tools.