How AI Saves Lives in Healthcare with Dr. Richard Milani, Sutter Health

[00:00:01] This podcast is produced by Outcomes Rocket, your healthcare exclusive digital marketing agency. Outcomes Rocket exists to help healthcare organizations like yours to maximize their impact and accelerate growth. Visit outcomesrocket.com or text us at 312-224-9945.

[00:00:25] Hello, everyone, and welcome to Health's The Beat Executive Speaker Series. I'm Dr. Lina Wen, and I'm here with Dr. Richard Milani, who is the Chief Clinical Innovation Officer at Sutter Health. Welcome to the conversation. It's great to be here. Thank you.

[00:00:48] Well, Sutter Health, I know, has just opened an innovation center, and you're also investing in and partnering with venture firms and early-stage digital health startups, which really seems like quite an interesting concept for a healthcare system to be investing in. So tell me more about the thinking behind the innovation center, and what are some examples of projects that you're the most excited about? Yeah, we have a lot of things on the plate. So, you know, we have the unfair advantage of being in the heart of Silicon Valley. We're in Northern California.

[00:01:16] Obviously, there's an awful lot of activity both in the VC space as well as in the startup space. And so we have the opportunity to interact with many of these on both VC and startup side and to identify things that could potentially help us as a health system. We know what our pain points are, and we're obviously looking for the ideal solutions that could help us. On our side of the fence, within innovation, not only is our job to be able to sort of monitor the community of new things that are taking place, but also build our own.

[00:01:46] And so we're doing a lot of internal work on not only artificial intelligence, but building clinical programs that sort of re-engineer how we think about healthcare. So give me a sense of what this looks like, as in, do you come up with an idea and then you find a company that's already doing this? Or do you look for companies that have interesting ideas to see how they fit into your system? Yeah, I think it's a little bit of both. I mean, we know where the pain points are.

[00:02:11] If it's something that we can build internally and we have a good deal of talent on board, then obviously that might be something we undertake ourselves. And likewise, you may be able to build two thirds of it and you need a partner to get to the last mile. And obviously we'll look for the right partner to do that. So we've done any combination of things. Clearly, we're working with a startup now that's involved in the mental health space. And again, there's parts of this that we can build and then we can co-develop. And we'll be launching that product, we hope, next year.

[00:02:40] Wow. So what are some other examples? You mentioned mental health, maybe in political administration, but what might be some other use cases? Yeah, well, one of the biggest pain points in health care, as you know, is documentation time, pajama time, the pain and suffering that caregivers have to give in terms of after hours work, as opposed to pure clinical work, which is actually patient facing. So that's where we partnered.

[00:03:02] I mean, we partnered with a couple of large documentation scribe type companies to be able to help us in terms of reducing that sort of documentation period. And we also have taken advantage of many of the AI tools that are embedded within our EMR that can help answer requests that patients are coming in with that we would then oversee. But it can reduce some of the time involved in creating the documentation piece.

[00:03:25] Is your sense that other health care systems are looking to you as an example and saying, look, we're identifying these pain points, these areas of need, and we're adjusting as well? Or is this what Sutter is doing that's quite unique? I think that the common pain points, most health systems are working towards the same goal, and we're no different than many others. There are other unique things, however, obviously, that each system may have an advantage of. I think one area that's really unique for us is back to the chronic disease programs.

[00:03:53] So we've designed and built programs that get two to three-fold better outcomes in terms of diabetes management, hypertension management, COPD management as examples, and standard of care. And if you're taking risks, so for instance, in a Medicare Advantage population where you're now the insurer, not only do you benefit in terms of providing better outcomes for your patients, but you also provide a financial ROI for your health system. And that's becoming increasingly difficult these days with reimbursement rates being cut.

[00:04:21] So we're looking for those natural places that we can provide a new way of providing a better clinical outcome that exists in a sort of a sporadic intermittent clinical visit twice a year. Well, there's something else I wanted to ask you about, because in addition to your role at Sutter, my understanding is that you're also an acclaimed health futurist. I would love to learn more about this and tell us about the kinds of forecasts you've made on the future of healthcare delivery. And also, what do you see as the future in the next five to 10 years in healthcare?

[00:04:48] Well, I don't know about the word acclaimed, but I can tell you that there's no doubt. The Institute of Medicine published an article in 2020, actually a paper, that described the amount of knowledge that is required to do the best clinical care for an individual. And the point is, is that human beings have a limited capacity to manage multiple different sort of options. We can only manage comfortably as humans, at least in clinical care, two to three options with one or two possible outcomes.

[00:05:18] Where we have crossed the line was in roughly 2020, nothing to do with the pandemic in terms of now new data that's available to clinicians where we have more than two possible outcomes and multiple choices. And it's beyond our capability. The moral of the story that the Institute of Medicine concluded is that there is no future for us without an AI assistant that's able to manage this complex set of data sources that are coming in and then narrow down the options to be able to be two or three.

[00:05:46] We don't have the compute capability internally to be able to do that. So there's no doubt in anybody's mind, if we're going to be successful in providing the best care that we can for human beings on a go forward basis, the tools will need to be further developed. But we'll need an AI assistant along the way to help us. What are other use cases for generative AI in health care? I know this is an area that you've written and talked extensively about. So you were mentioning AI as the digital assistant. Is it in diagnosis and treatment?

[00:06:15] Well, let me give you a great example of this. If you look at cardiac arrests or really rapid deterioration inside a hospital. So I know you're an emergency medicine physician, but now let's say they're up on the floor. Think about this. Human being is hooked up. They got monitors. They have nurses. They have doctors rounding all the time. And yet in spite of that, we still suffer for patients having a sudden code. If you have a cardiac arrest in the United States hospital today, the mortality inside the hospital is 75%. That's the national data.

[00:06:43] And all the research has been done in terms of resuscitation efforts, which have not yielded, not that we shouldn't continue, but they haven't yielded the benefits we'd hoped for. We built a model that's congesting the entire medical record on every single patient, updated every 15 minutes, which means a new lab result came in, a new vital result, new radiology result. And it was able to predict a rapid deterioration leading to death or an ICU within the next four hours. Talk about computational capability. There's no human being that could possibly do this.

[00:07:13] And then it provides an alert to a dedicated team that can go evaluate that patient and determine why the AI is sensing that this patient is going to have a sudden downturn. We were able to reduce our cardiac arrest and codes by 44% just by doing that. So that's a great example of where I need an assistant that can give me better capabilities that I currently have today in terms of just relying on our traditional system. I did some research last year looking into this, and I know that Kaiser has also implemented this in their Northern California hospitals.

[00:07:43] There was a New England Journal article also showing that this AI model reduces mortality, which is really, really significant. I mean, I think about, as you were saying, no human is going to be checking every lab result and looking at all the different pieces of data. And it seems like the synthesis of those data, the gathering of those data is a key place for AI. Absolutely. Not only is it the volume of data, but able to run a model and a compute to be able to calculate when a risk exceeds a threshold. That's beyond our capabilities.

[00:08:12] And we've done this for other disease states, even hospital-acquired infections. We can predict in advance who's the most susceptible to develop, say, C. diff in the hospital, and then design interventions so that that person never gets it, as opposed to trying to do it for everybody. So there's no doubt that we'll be able to provide better care for people than we do today by virtue of utilizing AI in the right way. I think there is a lot of fear around AI as well.

[00:08:37] I think once people hear about these use cases, as you've outlined, I mean, who could be against making medical care safer? Right. But maybe the fear is about chatbots somehow replacing the interaction between clinician and patient. Can you speak to that? And how do we make sure that in this digital transformation age that we don't lose the human connection that exists as well? Yeah, I think you bring up an excellent point. And you always want a human in the middle. So the AI, like going back to that example, can say, I'm very concerned about the risk of this individual.

[00:09:07] And now a human being with that special talent is going to go evaluate that individual and determine if that's correct or not, or why the AI is thinking that. And then do whatever they think is appropriate that somehow we've missed up until that point. So as long as we include a human in the middle, I think we're going to be safe. But we have to be cognizant of, you know, not letting these things go awry, make sure there's no bias, making sure that it's making the right decisions and updating it on a regular basis. So obviously, AI governance is critical.

[00:09:36] But we can't let this opportunity in terms of proving care for our patients to go unnoticed. What about other trends in health care, including home health or chronic disease management? Anything else that you as the health futurist, the acclaimed health futurist? Oh, yeah, acclaimed my... Well, there's no doubt that we're dealing with the largest influx of people aging into elderly, 65 plus in the history of the country. So we're now in what's called the peak zone.

[00:10:06] It started this year, 11,200 a day are aging into Medicare. And obviously, we have an aging population. And we have a small number, a reduced number of caregivers for that population. So the point is, there's not enough children to manage an elderly patient that might live alone or have health needs that go beyond sort of the routine going to the doctor a couple of times a year. That's an enormous problem. And we don't have the funds to be able to pay for home care for every single person in that situation.

[00:10:34] But there are ways that we can monitor people unobtrusively that monitor activities of daily living and other sort of things that can tell us when somebody is starting to slide. And then we can reach out. So the idea is, how can we use technology to identify who do I need to manage now, as opposed to trying to manage everybody to find the one person? Let the technology to point to us. Look at those two or three people there. And I think that'll be the way we'll have to design our future.

[00:11:00] I could ask you all day about the future, but we'll have to continue our conversation at the time. Dr. Milani, thank you so much for your time with us today. Thank you so much for inviting me.

[00:11:09] This podcast is produced by Outcomes Rocket, your healthcare exclusive digital marketing agency.

[00:11:40] Outcomes Rocket exists to help healthcare organizations like yours to maximize their impact and accelerate growth. Visit outcomesrocket.com or text us at 312-224-9945. Outcomes Rocket, your healthcare exclusive digital marketing agency.