AABB2025: Understanding Generative AI

Health care organizations are rapidly integrating artificial intelligence into their daily operations, trying to strike a balance between adopting these transformative technologies while still maintaining strict patient safety protocols and human oversight. The future is here, but how do we turn caution into acceptance and harness AI’s power?

An educational session, “Understanding Generative AI: A Friendly Introduction for Medical Professionals,” served as an AI 101 overview and underscored its beneficial role in the health care realm. One example shown was the National Marrow Donor Program (NMDP)’s new AI Center of Excellence, where senior AI strategy managers now oversee the integration of machine learning and large language models into registry operations. The organization is leveraging AI to analyze vast amounts of donor and patient data to optimize HLA matching and improve transplant outcomes at a faster pace than would be possible using traditional methods.

Other real-world applications of AI in health care are already in use, including:

• Medical imaging: Cytopathology departments are successfully using AI image recognition to identify cells, with human pathologists maintaining oversight.
• Patient communication: AI generates personalized donor education materials at specific reading levels and tones.
• Blood utilization: Real-time AI analysis helps reduce waste and optimize blood product usage.
• Educational content: Health care professionals create custom training materials, presentations and even gamified learning environments without needing design teams.

Speakers Ericka W. Narr and Chancey L Christenson emphasized that AI serves as a “productivity multiplier” rather than a replacement for human expertise. AI extends capabilities, allowing professionals to create sophisticated educational content, analyze complex data patterns and generate research materials that would previously require specialized teams.

Safeguarding patient health information (PHI) is paramount, and health care organizations are navigating strict privacy requirements by implementing enterprise-level AI solutions with business agreements to protect PHI. Microsoft Copilot and similar enterprise tools are becoming preferred alternatives to open-source models like ChatGPT for health care applications.

“I...really do believe that in five years from now [AI] will be woven into everything, whether we want it to or not. I think it’s just going in that direction,” said Narr. The transfusion medicine community is poised to embrace the challenge.