Hackensack Meridian earns first Joint Commission AI certification and says every tool gets traced like a hospital survey

Only a few months after the Joint Commission launched the Responsible Use of AI in Healthcare (RUAIH) certification, Hackensack Meridian Health has become the first recipient of the award.

The New Jersey-based health system leapt at the chance to put its AI governance processes to the test, said Joel Klein, MD, EVP and Chief Digital and Information Officer, continuing a longstanding tradition of pursuing thoughtful, strategic, and patient-centered digital maturity.

“Artificial intelligence has been part of our digital identity for years, long before the Joint Commission launched this new program,” he told Digital Health Insights. “We have always had a culture of constructive ambition, which is deeply rooted in governance.”

“When the Joint Commission decided it was time to have health systems stand up and declare what ‘good’ looks like, we wanted to raise our hand and participate. It just so happens that we completed the assessment first, but we’re proud to be at the forefront of benchmarking what we’re doing against ourselves and the industry, and hopefully providing some insight into what works well for industry at large.”

Following the AI threads from beginning to end-user

Joint Commission has a long history of scrutinizing processes and workflows within hospitals and health systems, traditionally to assess aspects of their safety and care quality. The new RUAIH certification takes a similar approach, said Klein, but applies those rigorous processes to AI tools, instead.

“The methodology they use is exactly the same as a hospital survey,” he explained. “It relies heavily on what they call ‘tracers,’ which is where they pick an AI tool out of the inventory and ask to see literally everything associated with it.”

“Our representative asked to see things like the contract with the vendor, the data we used to evaluate the tool, the processes we deployed to implement it, and how we conducted our risk stratification, among other things. They actually selected a contract with one of our biggest corporate partners, which was very long and complicated and had been negotiated extensively. It was actually a good choice, because it really showed how we moved through a partnership on the complex end of the spectrum.”

The assessor then spoke to multiple stakeholders, including users of the tool and asked detailed questions about how they were – or weren’t – using the product.

“It’s all about tracing that thread from start to finish,” Klein said. “Do they follow it slavishly? Do they not even know it exists? Does it help or hinder their daily operations? Talking to the folks on the ground is essential for understanding what responsible use looks like, as opposed to just responsible implementation.”

“One of our ER nurses, for example, told the team that they view the AI tool as another data point, just like the patient’s white blood cell count or their blood glucose, etc. They hang it on the whiteboard, so to speak, and take a step back and look at it in context with the other information they have about the patient. That’s a great way to view AI output and incorporate clinical tools into their decision-making process.”

“AI is everywhere” is no excuse not to govern it

Part of responsible AI use is knowing exactly what and where each AI tool is, and having appropriate governance surrounding each of them.

That’s increasingly difficult for most health systems, which are now gathering AI-enabled products at a breakneck pace, noted Klein. But it’s every health system’s responsibility to make sure that speed and volume don’t lead to a loss of control over the governance of AI tools.

“It’s true that AI is everywhere now,” he said. “But we still have to handle it with the appropriate attention to detail. It’s no different than saying PHI is everywhere. That’s true, as well. We’ve responded by treating PHI as an ‘everywhere’ problem. We have to do the same with AI.”

“That means making tough decisions about what to integrate and what to pass on. It also means having strategies in place to track what we’re bringing in, fit each tool into our governance framework, and continually monitor the risks and opportunities associated with our AI-powered toolkit.”

Developing robust governance that translates into responsible use

Creating a successful pathway for AI tools from vendor pitch to patient care requires health systems to invest in collaborative governance and structured risk assessment protocols, concluded Klein.

“You need to have a process, whatever it is,” he said. “And it needs to be inclusive of the people who have valuable input on the decision. That doesn’t mean you need 500 people at the table, but we work in organizations that have a lot of scientifically trained people, a lot of technically savvy people, and a lot of thoughtful people. Tap into that expertise so that you’re getting all of the most valuable perspectives from the beginning.”

“If you make sure that people understand the process, and get to be involved in the process, you can avoid the perception that these decisions are taking place in some smoky back room and affecting their job in some negative way. You want to be open and transparent so that governance becomes a shared responsibility, and appropriate use of these tools becomes a shared objective. That’s how you build a culture and a community that can take a leading position in AI as the care delivery environment continues to move in that direction.”

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].