Today in health, it we're going to take a look at a McKinsey article on generative AI in health care. My name is bill Russell. I'm a former CIO for a 16 hospital system and creator of this week health set of channels, dedicated to keeping health it staff current. And engaged. We want to thank our show sponsors who are investing in developing the next generation of health leaders, short test and artist site two great companies. Check them out at this week. Health.

, disparate sources of unstructured data, which bound in healthcare. Our now assets to power generative AI. Here's how private payers and healthcare systems can use technology to deliver care. And I'm going to skip a whole bunch of the narrative. And just come down to the bullet points. For you and it has private payers.

Should I go through privately. Here's the here's some of the things that private payers has, healthcare management. Synthesize clinical notes for care managers. , member services, Creek, , custom coverage summaries for specific benefits questions. This is actually good stuff. Provider relationship management, compare plan.

Product features and networks. , corporate functions generate HR self-serve functions, et cetera. , you know, the reason I like going to these things, I read, , anything that Accenture throws out, anything that McKinsey throws out anything that Deloitte throws out. , there's a couple others. I mean, you know, the specific healthcare ones as well, that are out there. The, , consultant pure-play consulting firms.

, I find them to be really good. , Thought. , pieces and they may not always be accurate for what I'm trying to do, or let me let Accurate's the wrong word. They may not always be applicable to my specific use case. However, I find them to be good to spur on my thinking around these things. There's a couple others under the private payers. I want to get to the healthcare.

Hospitals and physician groups, because that's where we live as a community. So here are the potential uses of generative AI in healthcare. According to McKinsey, number one continuity of care. Summarize discharge information and follow up needs for post-acute care, generate care summaries for referrals since.

Since the size specialist's notes for primary care physician team. So one of the things that generative AI does really well takes a lot of data, just a lot of, , unstructured data and it makes sense of it and it brings it back and puts it into a form that is readable and understandable. And in some cases, if we seen from some of the studies that we've looked at.

, more empathetic than you would get from a physician. So there you go. Continuity of care. Number two, quality and safety synthesize and recommend tailored risk considerations for patients based on their medical history and existing medical literature. It's interesting because again, AI can read all the medical literature. We can actually have it read the entire, all the medical journals that are being released. All the new studies that are being released.

And it can also then look through that entire medical record very quickly. Now, granted it's not going to med school and it will still require a physician to be on the other side of it, looking at it and determining the value. Of the information that is actually seeing. And R that the physician's actually seeing and, , and making determinations based on that. But think about the size of the medical record at this point, it is grown quite large.

Computers are going to be able to continue to take in more information. Then any individual human. And so

it can actually identify patterns. In the medical record and compare those against the medical literature and then create like a checklist for the physicians. Hey, take a look at these things. These are some of the things we found as we went through the medical record. And some cases. It, depending on the technology and how it lays it out, they could actually click on some of those recommendations and go deeper into the medical record. Did see those things that the AI considers relevant.

Let's keep going. Value-based care, improved documentation, accuracy, and leveraged, structured, and unstructured data to create patient education, videos, images, and summaries. Eh. , maybe, I mean, I could see that as a use case. I just don't see it as a high. Value use case at this point. , I don't, I don't see the economic model behind it.

, just yet. Network and market insights. Auto-generate provider segmentation, summaries by specialty summarize market performance and comparison based on external resources and data. That's interesting. Man, I would have to see that AI model. Inaction. 'cause I mean, at that point, I'm not sure we're talking about a large language model.

But we are talking about generative AI. So it would be interesting to see what kind of models they're talking about. And if those models can be trusted to take in those kinds of insights and provide details back at yeah, I mean, I could see it. I could see it. Reimbursement. Develop develop prior authorization documentation for payers generate a list of current conditions and potential codes based on voice electronic medical records.

Tax and other data, by the way, this is, this is already being done. And a lot of people are talking about the battle of AI. You're going to have the payer AI denying claims. You're going to have the resubmittal of claims by AI on the provider side. And you're just going to have AI systems bouncing these things back and forth hopefully quicker than we did in the past.

So even though it's bouncing back and forth between these systems, Hopefully it will be. A situation where you take the people involvement out of it. You have quicker cycles and things get processed a lot. , easier. If you could imagine, you know, you have a denial of claim, it comes back with the reason the, , AI system on this side digs into the medical record, comes back and says, here's the information that makes this a complete submittal. If it needs to go to the clinician for review before they submit it, they can do that.

Boom. It goes back. I would like to keep the clinician out of it. If we can. Because now you get, again, shorter cycles back and forth between denial of claims. To approve claims and that kind of stuff. So reimbursement, I think is an interesting use case and probably one that's supported with an economic model.

And when we will see pretty quickly clinical operations generate post visit summaries and instructions. Generate and synthesize care coordination notes changes in EMR dictations. And messages generate workflow, materials and schedules. Again, I think all of this stuff is going to be baked in. We're already seeing the notes be generated by the, ,

I guess that is AI. Now it didn't used to be just so you know, I mean, a lot of those dictation systems went to, , rooms either overseas or somewhere where you can get low cost labor. Who were essentially doing the dictation or reviewing the dictation and then moving it back in. So there was a lot of things that weren't AI before, but we're now seeing the new models are AI based type model. So you have the, you have the notes being generated that way. And then clearly you could take multiple notes and you can start to create summaries and instruction sets and those kinds of things.

Clinical operations. I think it's going to be a slow. Grind to see this come through. , I think, but I think a three-year time horizon on that three to five-year time horizon on that we will see a significant amount of clinical operations. Being done with AI. , corporate functions. It developed code assist, cybersecurity test case generation.

And quality assurance. Absolutely. , procurement draft RFPs. Absolutely. I mean, all of this stuff, all this stuff that is not related to life and death situations within the healthcare system, I think are prime candidates for this to be utilized, you will see it. And, , all areas of it, I think quicker than anywhere else.

Within the organization, if not then, , maybe your it organization is not forward-leaning. Anyway, clinical analytics leverage conversational language to obtain analytics. Insights use AI assisted coding to automate repetitive tasks to generate new code. Sure. , you know, I mean, I can see that, ,

Man, there's going to be so many use cases coming out in the next couple of years. It's just going to be everywhere. People are already tired of hearing the word generative. AI are a little skeptical around it. And probably for good reason, it is at the peak of the hype cycle, but the reason it continues to be at the peak of the hype cycle. And I keep coming back to this says it's real.

If it wasn't real, it wouldn't be at the peak of the hype cycle. It would be like, say, , you know, , the internet. Oh, we're tired of hearing about the internet. Well, it continues to change things even today. The fact that we have this, this, , incredible network of computers and systems and devices around the world continues to change how the world operates.

, every day, since the first time people are saying, Hey, it's at the peak of the hype cycle. I think AI is in that same, same boat at this point. , and then they have consumer as well, analyze customer feedback by summarizing extracting themes, online text messages and whatnot. We're already seeing AI applied in call centers. And if you're not applying AI in your call centers right now, they are, ,

They are not efficient. They are not at the top of their game and they're not leveraging the things that they could be. , I like this article and I recommend this to tackling healthcare's biggest firms with. Generative AI. I recommend it. The end of this article is very good. I'm not going to have time to go through it all, but, , bringing gen AI to healthcare, they talk about evaluating the landscape, sizing up the data. The data's important extracting the greatest value from the gen AI opportunity will require broad, high quality data sets.

And a lot of cases, your house, some may or may not have that. You're going to have to look at what types of data sets are good for training these models. And, , understand how you are going to get your data in a form that can be used to train these models and then determine what types of models this is, why we're evaluating the landscape, what types of models you're going to use. Again, you're not going to have one size fits all within your health system.

There should be multiple AI models within your health system. He goes on address risks and bias with your models, invest in people in partnerships. And

I'll just give you a little taste of this bringing gen I. To healthcare organizations will affect not only how work is done, but by whom it is done. Healthcare professionals will see their roles evolve. As the technology helps streamline some of their work, a human in the loop approach, therefore will be critical, even though many processes may fundamentally change. And how someone does their work may look different. People will still be critical to all areas, touched by gen AI to help bring these changes to healthcare.

Organizations must learn how to use gen AI platforms, evaluate recommendations and intervene when the inevitable errors occur. In other words, AI should augment operations rather. Other than replace them. So you get that idea of the ongoing partnership. Again, highly recommend this article. If you get a chance it's out on the McKinsey and company website.

To develop the next generation of health leaders, short test and 📍 artists. I check them out this week. health.com/today. Thanks for listening. That's all for now.