Craig Kwiatkowski, senior vice president and CIO of Cedars-Sinai, and his team have been advancing the well-known health system in the realm of artificial intelligence.
Their most notable accomplishment is Cedars-Sinai Connect, an AI-powered virtual primary care app. The app, released in fall 2023, has already increased primary care capacity by 11%. This equates to him building three new clinics. From San Diego to Sacramento, he has helped more than 6,900 patients receive care through his more than 9,200 virtual visits.
But that's just one of the achievements of AI. Kwiatkowski, who holds a doctorate in pharmacy, is immersed in this technology and how it can improve outcomes, enhance the patient and provider experience, and reduce clinician burnout. is finding.
The following is the eighth interview in our series about top voices in the healthcare IT industry discussing AI. In this Part 1, Kwiatkowski shares his thoughts on hot topics. AI in general healthcare. In Part 2 tomorrow, we'll discuss the AI project at Cedars-Sinai.
Q. How do you decide whether to build or buy AI? And what are the biggest challenges in integrating AI technology into existing systems?
A. The question of whether to build or buy depends on the situation. We start by looking at each with fresh eyes and understanding the problem we're trying to solve. And when possible, we want to find solutions within existing systems and platforms.
So we're the Oracle shop for ERP and the Epic shop for EHR. If workflow is enabled on one of these platforms, you will typically use that functionality whenever possible. And those vendors and other big vendors have built great roadmaps of tools, and we want to leverage those roadmaps as much as possible.
An example is Epic. We are starting to use in-basket message response technology. This is also used by other companies, and drafts are queued for doctors to further edit and send.
Another feature we're starting to work on is graph summarization. It uses AI to categorize all the information in the graph and launch a course note or, in some cases, a discharge note. Among these vendor solutions, what we're trying to solve is physician burnout and health.
If one of these platforms or solutions doesn't satisfactorily solve your problem, ask other vendors to support your solution and consider whether to build something yourself. There are obviously many variables that go into that decision, perhaps too many to discuss here, but thematically, the key variables there are resource capacity and availability, feasibility, and efficiency. It will be the center.
In some cases, you can also purchase ways to accelerate functionality. Ambient document solutions are a good example. We're not going to try to build it ourselves. That would be too difficult, time consuming and expensive. But sometimes we're willing to invest and build if we recognize a gap in the market or if it makes sense to do it ourselves.
DIY gives you a little more control and flexibility to create something tailored to your specific needs. We'd like to think, and people in healthcare would like to think, that we're reaching a point where healthcare is standardized and scalable, but there are bespoke workflows and processes that don't always work well. There are still many. Buy it, or even if you do, you'll spend a lot of time configuring it locally.
So there will probably always be some trade-off.
The second part of the question was integration and workflow. I alluded to it, but the more you can incorporate these tools into your existing workflow, the better. Frankly, from a physician's perspective, removing them from the physician's workflow is a non-starter. No separate logins, no side-by-side portals or dueling screens, no extra keystrokes, and ideally fewer keystrokes.
This, in turn, leads to successful integration and deployment of these solutions in the short and long term.
And sometimes these solutions are just what you need, whether you like it or not Workflow redesign. In some cases, there is no way around it. So in such cases, we need to have the right 'why' story and all the necessary change management in place to support those who will be most affected on the front lines.
Q. How are patients and clinicians responding to these tools, and is there anything more healthcare organizations can do to encourage use or expand adoption?
A. We have been using AI for many years, primarily in the area of rule-based machine learning. We've had a lot of success incorporating these types of tools within our workflows and have successfully deployed them for all kinds of use cases, including patient risk prediction. , deterioration, hospital capacity, and patient flow.
But when it comes to new things; Generative AI tools are still in their infancy. I think, like most areas across the country, we're taking a cautious approach to testing these tools and making sure they're safe and effective. We focus on organizing our approach around his FAVES principles: Fair, Appropriate, Valid, Effective and Safe. And make sure you understand how these tools work and how they work in the short and long term.
There's still a lot to learn at this early stage, so to speak. And we're investigating whether these tools work. And some examples I mentioned. Is the in-basket note draft complete? Are keywords missing? Where can context be lost? Where is additional information inserted that wasn't there in the first place? Hallucinations is something that people are aware of and want to be aware of.
We're intentionally taking a slightly restrained and cautious approach to deployment to ensure we deploy our solutions safely and responsibly. This is actually helpful from an implementation perspective as well.
As you build trust in your early adopter user community, they turn into evangelists who help share stories about how these tools work, how they can help, and where the opportunities lie. can do. It's very helpful to have friends with whom you can share that information. Word of mouth has proven to be a very powerful tool for accelerating or slowing down adoption.
However, there is certainly a lot of interest in these tools and excitement about their potential. Frankly, I don't see any silver bullets that will solve many of the challenges we face in the short term, but these tools, generative tools, have incredible potential in the medium to long term. . Therefore, it will be fun to solve them.
Editor's note: This is the eighth installment in a series featuring top voices in the health IT field discussing the use of artificial intelligence in healthcare. To read our first feature on Mayo Clinic's Dr. John Halamka, please visit click here. Click here to read his second interview with Geisinger's Dr. Arupen Patel. Click here to read his third installment by Meditech's Helen Waters. Click here to read his fourth installment by Sumit Rana on Epic. Click here to read Part 5 of General Brigham Mass by Dr. Rebecca G. Michelis. Click here to read his sixth installment by Dr. Melek Somai of Froedtert & Medical College of Wisconsin Health Network. Click here to read Part 7 by Dr. Brian Hasselfeld of Johns Hopkins University School of Medicine.
Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email: bsiwicki@himss.org
Healthcare IT News is a publication of HIMSS Media.
