The Department of Biomedical Informatics brought together 19 clinicians for an intensive two-day workshop focused on building, evaluating, and responsibly applying generative AI in clinical care. On August 4 and 5, 2026, the University of Utah Department of Biomedical Informatics (DBMI) hosted Generative AI for Clinical Care: Build, Evaluate, Apply, a hands-on workshop designed to give frontline clinicians practical experience developing and evaluating generative artificial intelligence tools for healthcare.
Nineteen clinicians participated in the two-day program, working alongside DBMI faculty mentors and DBMI student coaches to explore how generative AI can be applied to real-world clinical challenges.
Rather than beginning with the technology, the workshop challenged participants to begin with a problem from clinical practice and determine where, and whether, AI could provide meaningful value.
Building AI Around Clinical Needs
The workshop combined foundational instruction with guided, hands-on development. During the first day, participants explored the fundamentals of modern AI, designed a focused clinical use case, and began building the components of their own AI tools. The second day shifted into an applied clinical AI lab, where participants developed prototypes, tested and refined their approaches, and shared their work for feedback.
Participants also received hands-on training with AI Workbench, an AI development tool incorporated into Epic. The training gave clinicians an opportunity to explore AI development within the context of the clinical environment and consider how tools can be designed with existing healthcare workflows and systems in mind.
The program was designed to take participants beyond simply creating a working prototype. Clinicians were encouraged to consider the questions that ultimately determine whether an AI application belongs in healthcare: Is it useful? Is it safe? How should it be evaluated? Does it fit within clinical workflows? And does it provide meaningful value to patients and clinicians?
Participants were also introduced to considerations for moving AI tools beyond isolated applications and toward real-world healthcare environments, including cross-platform approaches, AI-native intelligence layers, and FHIR-based integration.
Learning Through Collaboration
DBMI faculty mentors Yves Lussier, MD, Carolyn Scheese, DNP, MS, RN, and John Thomas (JT) Menchaca, MD, brought expertise spanning clinical AI, biomedical informatics, clinical practice, interoperability, and implementation to the workshop. David El-hata and Christian Mahoney, members of Kensaku Kawamoto, MD, PhD, MHS’s team, also contributed their expertise and supported participants during the workshop.
The program was built around a clinician-for-clinician model, pairing that expertise with the firsthand knowledge participants brought from their own clinical environments.
DBMI students Robert Barber, David Villarreal-Zegarra, Liam Wilson, and Jacqui Wright served as student coaches, providing additional hands-on support as participants developed their projects. One student coach reflected that some of the most valuable learning came from conversations with clinicians about the problems they were trying to solve and from watching those ideas develop into prototypes. The experience reinforced an important principle for clinical AI development: the starting point should be the clinical problem, not the technology.
Moving From Prototype to Practice
By the end of the workshop, participants had worked through the process of defining a clinical use case, developing an AI prototype, evaluating its performance and safety, and considering a pathway toward implementation.
That progression reflects a larger challenge facing healthcare as generative AI tools become increasingly accessible. Building a prototype is becoming easier. Determining whether that prototype is sufficiently useful, safe, validated, and integrated to support clinical care remains considerably harder.
For DBMI, that gap represents an important opportunity for biomedical informatics education.
By bringing together clinicians, informaticians, faculty, and students in a hands-on environment, the workshop demonstrated how DBMI can help healthcare professionals move beyond simply experimenting with generative AI toward rigorous, clinically grounded and responsible development and implementation.