Q&A: What Is the Relationship Between AI and Clinical Informatics?

HEALTHTECH: How does clinical informatics improve patient and provider experiences?

BEENE: First, I want to highlight that at Trinity Health, we call our team health informatics. In my professional experience, clinical informatics limits the impact of the profession to only clinical activities, whereas health informatics is comprehensive and denotes more of a continuum. That means our team here supports health informatics for the continuum of care and all interactions that occur within that environment, from outpatient offices to acute care settings to long-term care such as skilled nursing facilities.

HEALTHTECH: That’s an interesting distinction. With that in mind, how does health informatics support the continuum of care?

BEENE: Through health IT tools that support those environments of care. For example, you have a patient in your medical group, and you have tools in that environment that clinicians and front-end colleagues can use to help facilitate care workflows. Health informaticists are those professionals who assist clinicians and front-end colleagues with using the tools.

We collaborate with colleagues intentionally to design workflows. Both entities collaborate to make that work. It cannot just be the informaticians who help to develop a workflow that’s going to make your processes more efficient. You must be a partner in helping them design that workflow. That’s one of the major ways to encourage adoption of health IT.

DISCOVER: Clinical workflow optimization creates better patient outcomes.

HEALTHTECH: What are some of the main challenges healthcare informaticists face today?

BEENE: Overall, I would say, rapidly evolving innovation. In healthcare, there’s a push and pull. There are all these things that we could engage in, so we’re mindful of some of those pitfalls and some of that misalignment.

When you think about a rapidly changing environment, our workforce has to remain on that cutting edge, but it’s difficult. For example, with AI, in communicating the real promise of artificial intelligence, you can say it’s great, but it doesn't replace the politics and marketing of what AI says it is or could do within an organization. We try to balance an eagerness to dive into all the latest AI tools, while taking a thoughtful approach to examining which ones align with our strategies, goals and who we are as a health system.   

As an informatician, you have to have the latest knowledge to explain the contextual realities of how AI is really going to work. Sometimes, that conversation gets challenging when there is a buzz around the “next best thing.” However, as an informaticist, our role in that moment is to be an adviser and a consultant. We have to be able explain the realities of what it is and what it’s not. You don’t want to miscommunicate the capabilities of any health IT, including AI, to where clinicians have unrealistic expectations. You want to be sure they have a clear understanding of all aspects of a tool.

But there are opportunities with AI in healthcare as well, and we are on the precipice of understanding what that means for patient outcomes. Can we get algorithms that will support the identification of populations? We can. Can we provide operational efficiencies with HR AI? Of course. We want to explore how AI can mitigate the administrative burden of documentation, for example, but we have to make sure that we understand — especially on the clinical side — that there’s a patient on the other end of that process. We have to ensure they’re safe and that quality care is being delivered.

It’s important to approach vendors thoughtfully. They bring a lot of marketing data and information to the conversation, and if you’re not well versed in those research areas, you may not be aware that the data is being skewed to support a certain narrative. It’s a complicated space for the health informatician. As a leader, I try to stay engaged so I can be that adviser and consultant, not only for the stakeholders who come to me but also for my team to ensure that they have the wraparound services and guardrails needed for AI use cases.

HEALTHTECH: How else is AI impacting the field of clinical informatics?

BEENE: There are other business areas where AI can impact the workforce. For example, can we adopt AI tools to help us be a better workforce, such as evaluation tools after we’ve implemented new technology to ensure that our end users and our stakeholders are getting the services they signed up for or purchased? Yes, but that’s to come. Can we be smarter in how we look at clinical data? Yes. Can AI be the tool to help us? Yes. So, I would say the impact of AI is probably something that is to come. How can we start to tangibly define the true impact? I think we’re still learning and evolving, honestly, so we don’t fully know what all the impacts are just yet.

HEALTHTECH: What do clinical informaticists need to do to prepare for AI adoption?

BEENE: In our team, we have approached the consideration of AI tools just as we would any other health IT tool. We have a process for each one of these engagements. To determine what we focus on, we engage with the end users, whoever they may be.

For example, let’s just talk about our clinician stakeholders: We would prepare them to ensure that they understood what their involvement would need to be with their workflows. It always comes back to the workflow, because to me, it’s like walking and talking. For a workflow to happen, you don’t just look at the workflow and say, “Make it happen.” There has to be a natural engagement with the tool to make it better.

Think of your life before the smartphone. Now think of your life with and without it. There are nuances to both, and each person uses their smartphone differently. So, I’m going to make that tool work for me in my workflow as you make your tool, whatever it is, work for you within your workflow. You’ll use the apps that you need to make your day better while I use the apps I need to make my day better. We have to adopt AI into those workflows in a way that’s going to make the clinician most effective and efficient. We have to ensure that they’re going to find value in using that tool.

We wouldn’t change our implementation approach just because it’s AI. It’s still something that has to be incorporated daily by our end users. We wouldn’t change that approach at all.

One thing I want to reinforce, though, is that stakeholder or end-user engagement is critical to the success of any tool — AI, electronic health record or what have you. But no matter what is being incorporated into an environment, adoption should be based on collaboration between us, as your health informatics professionals, and you as the end user. We have to work together.

Click the banner below to sign up for HealthTech’s weekly newsletter.

HEALTHTECH: From a clinical informatics perspective, where should AI investments be focused?

BEENE: At Trinity Health, we are looking at this through the lens of a very large healthcare system, with several leaders across the system. Essentially, we’re a “system of systems,” and so we have to take a broader, more comprehensive view as we consider how we approach investments.

We are implementing and evaluating several tools and applications that focus on patients, automation, and clinician and operational efficiency — with an emphasis on the back-office area where there is lower risk, higher feasibility and the ability to deliver the most value.

Trinity Health has an AI governance group, which I am a member of, that includes stakeholders from across our system, such as CEOs, CMOs from our Regional Health Ministries, CIOs and others who can represent multiple perspectives regarding AI. We have to be thoughtful because AI is always evolving. While we have got to keep our finger on the pulse of it, we must keep an ethical lens on it too, aligning to our core values, always keeping our patients first and mitigating risk.

I’ll share a tactical example: We’re implementing a few AI tools for our revenue area — where we know there is a lower risk to our organization — to provide efficiency and make things more accurate. The areas where we tend to be more cautious are the clinical areas because you’re impacting patients. We must ensure that those tools are safe to deliver that care to those patients, in an ethical manner.

READ MORE: Healthcare organizations need AI data governance strategies that ensure success.

HEALTHTECH: What lessons could healthcare leaders take from the field of clinical informatics as they prepare their data for AI?

BEENE: Communicate the true possibilities of what the AI tool can do. That is one of the early lessons we learned about health IT overall: Be transparent about what it can and cannot do. I know there is a lot of pressure to push the promise of AI — and for some organizations, I totally respect that — but when we overpromise and underdeliver with the financial investment, then you have something to answer for later. Communicate plainly, honestly and with transparency about what you know and what you don't know.

The organization should also be prepared for the change that’s coming with that AI tool. It’s about making sure everyone understands what exactly is coming and how they’re going to prepare for it in their environment. That’s so important, because organizations will invest in something that they think is going to bring them certain promise, but it ends up being underutilized. Why? Because there was a lack of understanding of the problem the organization was trying to solve. How do you make this tool valuable in my daily activities? There must be organizational buy in.

Think about your daily life. If you don’t think something is valuable, you don’t go to that place, or you don’t invest your dollars there. People have to understand how the tool is valuable to them if they’re going to use it.

As informaticists, we have to make that connection real. What leaders can take from us, in the implementation of any health IT tool, is that they have to make it make sense to the people who are using it. They have to adopt it. They have to see that it brings value to them and is not a burden. That takes a lot more than just communication. So, be fluid. Be open. Be adaptive.

There’s so much fluidity here that it’s almost like, if you dip your toe, you have to make sure that you’re only dipping your toe, not your entire foot, because there’s something else that’s just around the bend that’s going to be better. It may be delayed gratification, and you have to be OK with that. You have to be OK with the fact that you just invested by dipping your toe versus waiting to dip your whole foot.

Organizations can’t always pivot that quickly. These are multimillion-dollar investments. Sometimes you have to take the information you have and make a decision based on what you know now. It’s easy to overanalyze. It could put you in a space of analysis paralysis, which could become a barrier to advancing any type of investment in AI. It all moves really fast. You must know whether or not you’re a risk-taker, what part of the risk continuum you and your organization are on, and where you feel that it’s OK to do that. Then jump in.