Skytron Blog

Most of us have watched a behind-the-scenes show explaining how something is made. This newsletter, we’re taking that idea in a different direction. Instead of looking at how AI is being used in healthcare, we’re looking at something most people never see.
 

How does a hospital decide an AI technology is ready for patient care?

A sales representative doesn’t walk into a hospital, demonstrate an AI solution, and leave with a purchase order. Like any technology that could influence patient care, AI-enabled solutions go through an evaluation process before they become part of a clinical environment.

The process is designed to review the available evidence, understand potential risks, determine who should be involved in the decision, and establish how the technology will be monitored after implementation.

That process is becoming more defined. In June 2026, AORN released its first evidence-based Guideline for Integration of Artificial Intelligence in Perioperative Practice.

The decisions aren’t left to a single department; the recommendations encourage participation from clinical staff, information technology, cybersecurity, legal, compliance, quality, and organizational leadership.

The guidelines describe governance as an ongoing process that begins before implementation and continues throughout a technology’s lifecycle. In practice, that means evaluating the available evidence, identifying appropriate stakeholders, establishing oversight, educating end users, and continuing to assess performance after the technology becomes part of clinical practice.

An aspect that stands out is who should be involved in the evaluation process. Rather than leaving these decisions to a single department, the recommendations encourage participation from clinical staff, information technology, cybersecurity, legal, compliance, quality, and organizational leadership. That reflects the reality that technologies capable of influencing patient care often have implications that extend well beyond a single clinical area.
 

What It Looks Like

One organization that has publicly shared its approach is the University of Wisconsin Health.

As more predictive models became available across the health system, leaders recognized that evaluating each application independently wasn’t sustainable. Instead, they developed a formal governance structure designed to provide consistency while allowing flexibility as adoption continued to grow.

UW–Madison began by identifying governance challenges and defining guiding principles. From there, it established an institutional oversight committee responsible for setting standards across the organization. Instead of that committee evaluating every application itself, smaller

working groups were created to review individual technologies based on their intended clinical use. In other words, one committee established the framework, while smaller groups evaluated each application where it would actually be used.

The governance model was designed to grow alongside the organization’s use of these technologies. As new technologies were introduced, the oversight process could expand without starting from scratch each time.

While every health system will develop its own approach, examples like this offer a glimpse into the work that happens long before a new technology reaches the bedside or operating room.

AORN’s guideline is a first step. It will be interesting to see what this process looks like five years from now.
 

 
Resources
1. AORN. Guideline for Integration of Artificial Intelligence in Perioperative Practice.
2. University of Wisconsin Health / UW School of Medicine and Public Health. Governance of Clinical AI.
Journal of the American Medical Informatics Association. 2026.
3. AORN. Guideline Summary: Integration of Artificial Intelligence in Perioperative Practice.