Transcript
Speaker: I'm your host, Jenny Owens.
Speaker: And in this episode, we're going to talk about a book that I read recently called Hacking Healthcare by Tom Lowry.
Speaker: I've seen a few of Tom Lowry's columns around.
Speaker: He's written in a bunch of different publications about artificial intelligence and healthcare.
Speaker: And I thought it would be really interesting to grab one of his longer form publications and see how he develops some of these ideas.
Speaker: I thought this was interesting.
Speaker: Hacking Healthcare is kind of a compendium of a bunch of articles that he's written.
Speaker: pulled together, given some framework and introduction by Eric Topol, which is also interesting, and given kind of a narrative arc over the course of the book.
Speaker: I wanted to specifically focus, first of all, on the book overall, and then I wanted to dive in deep on chapter 16, which is on AI-driven leadership.
Speaker: So overall, the book is good.
Speaker: It was an interesting read.
Speaker: I'm glad that I read it.
Speaker: I do find that the book is a little...
Speaker: Fluffy on specifics.
Speaker: So there's a lot of, well, AI will solve this, or health organizations are already using AI to do this or that.
Speaker: They're using AI to predict no-show rates.
Speaker: They're using AI to do this or that.
Speaker: And there's not a lot of concrete grounding points into, okay, if health organizations are using AI to predict no-show rates,
Speaker: Is that working for them?
Speaker: Is it better than the existing systems?
Speaker: Please give us some metrics.
Speaker: Help me help me really pin down how AI is improving the health care system.
Speaker: The book is specifically billed as how AI and the intelligence revolution will reboot an ailing system.
Speaker: And the overall, the book seems a little bit unparalleled.
Speaker: Long on what AI can do, which is good, right?
Speaker: AI can do a lot of things.
Speaker: We're still in the very early stages.
Speaker: A little bit maybe more interested in where AI can take us and less interested in asking questions about how will it get us there?
Speaker: How will the intermediate stages look?
Speaker: What does implementation look like?
Speaker: Overall, I did think it was an interesting read, definitely a worthwhile read, but it raised a lot of questions for me, which is always the hallmark of a good book.
Speaker: If I find myself thinking, wanting to engage and ask a lot of questions, for me, that's a really good read.
Speaker: I wanted to dive in primarily on chapter 16, which is one of the most interesting chapters for me.
Speaker: on AI driven leadership.
Speaker: This is something that I haven't really thought about because my focus has been more on, okay, we have a given AI technology, we have a defined clinical or operational problem that we want to use.
Speaker: How do we take this peg and put it into this hole?
Speaker: How do we determine the goodness of the fit?
Speaker: This chapter really dovetails very nicely with the previous book that I read on business strategy, looking at AI driven leadership and what does it mean to really have AI driven leadership at the strategic level?
Speaker: So one of the quotes that I thought was really fantastic was he says, success, especially at scale, can elude even those making sizable investments in AI.
Speaker: So just putting a lot of dollars or a lot of effort into AI is not enough to get you to a successful place.
Speaker: You have to have a really high tolerance for iteration and for gradient based learning, which is not something really that I've found healthcare systems have frequently for tech.
Speaker: We wanted to put a solution in.
Speaker: We want it to work flawlessly right away.
Speaker: We want it to be head and shoulders above the previous solution.
Speaker: And we really do need to have a greater tolerance for systems that learn and develop and grow.
Speaker: I went off on a bit of a tangent here, wondering if we could send new tech off to like med school or an internship or residency or fellowship.
Speaker: Like, is there some sort of
Speaker: holding tank or sandbox where we can put this new technology to train it up so that we are not actually doing drills with live ammunition in our fully live ecosystem with live patients.
Speaker: This is a subject that is worth a lot more thought, but I did want to raise it here.
Speaker: One of the ways in which this particular chapter dovetails really nicely with our previous book, the new Lords of Strategy, is in thinking about value.
Speaker: AI-driven leadership requires a focus on metrics and value that are not just financial.
Speaker: When I was reading the strategy book, I really wrestled with the concept that the purpose of an organization or the purpose of a company is to make money for its shareholders.
Speaker: not to produce a product that customers want to buy, not to benefit the community, but to produce value for its shareholders.
Speaker: And I found myself wondering, what does this mean when you're working in a nonprofit health care system?
Speaker: What is the purpose of a nonprofit health care system?
Speaker: And I don't necessarily think that it's different.
Speaker: We just have to be careful about how we define value and how we define shareholders.
Speaker: When you're a nonprofit health care system, value is not just dollars.
Speaker: Right.
Speaker: Value is overall health.
Speaker: Value is community benefit.
Speaker: Value is the health of your employees and what you're doing for the overall community in which your hospital is situated.
Speaker: And your shareholders are not people who own a chunk of your company.
Speaker: I'm being very fast and loose with the definition of shareholders here.
Speaker: It is all people who might walk through your doors, right?
Speaker: So everything that you do as a nonprofit health system that can increase that value, not just monetary value, but health value, economic value for your shareholders, who is everybody who works in your health system and everybody who might walk through your doors as a patient.
Speaker: And we're all patients someday.
Speaker: If we're not now, we all will be patients someday.
Speaker: So if you think about value in broader terms than just financial, thinking about AI driven leadership, having an eye on metrics that are not just financial value, thinking about, does this give us operational efficiencies?
Speaker: Does this allow us to see patients sooner?
Speaker: Does it allow us to see patients closer in time to when they think, hey, I want an appointment?
Speaker: Does this eliminate some of the burdens that we have on our caregivers that are not directly in service of their serving the patient, right?
Speaker: Does it eliminate our documentation burdens?
Speaker: Does it clean up some of our financial workflows?
Speaker: I thought this was a really interesting connection between the two books that I read recently.
Speaker: And I love finding little through lines like this because this is really helpful.
Speaker: Tom mentions that AI driven leadership requires a few other things as well.
Speaker: One of them is an absolute crystal clear vision of where we want to be right as a health system, as a field in general.
Speaker: What is where do we want to get to?
Speaker: And then a high degree of flexibility about how we get there.
Speaker: Right.
Speaker: If the goal is.
Speaker: to have doctors and nurses doing what doctors and nurses do best, and then to have computers doing as much of everything else as possible.
Speaker: Great.
Speaker: That's a crystal clear vision.
Speaker: I'm not saying it's a correct vision, but that is a clear vision.
Speaker: Let's think really creatively about how we get there.
Speaker: Yes, let's do pilots.
Speaker: Let's try things.
Speaker: But then we need to move past pilots and proofs of concept and put into implementation the things that are actually working and performing well.
Speaker: And in order to do that, you have to choose your pilot metrics very carefully.
Speaker: Tom also calls out that we want to, that AI-based leadership really needs to have a focus on patient and provider experience and that you really need to be evangelical about what AI is doing and can do within your healthcare system.
Speaker: Part of that clarity of vision is really understanding, okay, where are humans and where is AI?
Speaker: Not just, you know, what are we doing for patients, right?
Speaker: Where are our doctors?
Speaker: Where are our nurses?
Speaker: Where are our
Speaker: food services folks, our EVS, what are the things that absolutely you 100% need a human to do?
Speaker: But also in our AI systems, where are the humans in the loop?
Speaker: Is the goal to have completely autonomous systems?
Speaker: Is the goal to have a human in the loop as the first point of failure?
Speaker: You fail over to a person immediately.
Speaker: Having a very clear vision about what in your AI system fails to a human and at what point will be very helpful.
Speaker: I think it's really interesting to think about our tolerances for learning, for iteration and for gradient based learning, because this is the way that many AI systems work, right?
Speaker: They are iterative.
Speaker: They need to be trained on your data in your health system.
Speaker: And I think it is helpful to think about the approach that we feel most comfortable with as a health care system.
Speaker: highly supervised learning, reinforcement-based learning.
Speaker: That requires an awful lot of human effort and human involvement in the tech.
Speaker: It's also maybe not necessarily the best way to get the best
Speaker: results out of your artificial intelligence system.
Speaker: So I think it is worth investigating where do we want really humans in the loop?
Speaker: Where do we want humans to be training?
Speaker: And at what points are we comfortable saying, okay, we would like this to be autonomous and we will be checking in at regular points to ensure that our AI has not generated results that are unacceptable to us for various reasons.
Speaker: So overall, Hacking Healthcare, interesting book, lots of great thoughts on AI-driven leadership and the clarity of the vision, the flexibility on how we get there, moving past pilots and into actually putting things into practice and thinking really critically about what is our tolerance for iterative learning and where is the appropriate role of humans in both sides of this equation.
Speaker: This is really helpful.
