April 3, 2023: Charles Boicey, Chief Innovation Officer at Clearsense joins Bill for the news. Chat-GPT is the key word of the ViVE 2023 conference. How does the development of language models such as GPT-3 and GPT-4 impact the healthcare industry, and what potential limitations and benefits do they offer? How has the development of Chat-GPT 4 impacted call centers, especially in relation to the use of artificial agents and language generation? What are the advantages of using Chat-GPT 4 for information retrieval, and how does it compare to other similar technologies such as Siri and Alexa? How has Microsoft been able to gain an advantage over Google in the field of natural language processing, and how might this impact the industry in the future? What are some potential applications of CCPT four for data retrieval and interrogation, and how might it evolve over the next decade?
We understand that staying ahead of the curve regarding Security Priorities can be challenging. Join us, April 6, 1:00pm, for this webinar to learn how CISOs in healthcare address Security Priorities for 2023 – insights that can help keep your healthcare organization safe and secure. https://thisweekhealth.com/ciso-priorities-2023/
Alex’s Lemonade Stand: Foundation for Childhood Cancer Donate
This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.
Today on This Week Health.
Nothing in the history of mankind from a technology. And or from an application perspective has ever gone from zero to a hundred million users in two months. That's something to think about. (Intro)
Welcome to Newsday A this week Health Newsroom Show. My name is Bill Russell. I'm a former C I O for a 16 hospital system and creator of this week health, A set of channels dedicated to keeping health IT staff current and engaged. For five years we've been making podcasts that amplify great thinking to propel healthcare forward.
Special thanks to our Newsday show partners and we have a lot of 'em this year, which I am really excited about. Cedar Sinai Accelerator. Clear sense crowd strike. Digital scientists, optimum Healthcare it, pure Storage Shore Test, Tao Site, Lumion and VMware. We appreciate them investing in our mission to develop the next generation of health leaders.
Now onto the show.
(Main) 📍 All right, here we go. We're doing a Newsday episode from Vibe here in Nashville, Tennessee, and we are joined by Charles Boise, chief Innovation Officer for Clear sense. Yeah. Charles, always good to sit down with you. Yeah, great to be
here. Bill, what?
In our booth? Yeah. In your booth. And it's newly designed booth.
Yeah. The marketing team did a good job. Looks good. Alright, so here's what we're gonna do. We're going. Count the number of times G P T four gets said on this episode. And for every time it gets said we're gonna do like punch buggy. Yeah, go ahead and hit 'em. But we'll also throw a dollar towards the Alex Lemonade stand.
What about chat? GTP 3.5. Same rule applies. There's
there's $1. Yeah. We'll just keep counting them. 3.5 counts. Okay. It's amazing how many just normal conversations it's coming up and I do want to talk about that. I do wanna talk about the conference a little bit. Let's start with this.
I just got off the stage with AWS and we talked about cloud and. And how cloud brings agility and whatnot to the industry. You're gonna be on stage tomorrow?
Yeah. Tomorrow. Yeah. We'll talk about trend versus trendy. Ooh, I love that game with you. So, flying cars and healthcare,
flying cars is
trendy.
Flying cars are trendy. They've come up every 10 years since 19. All right. And what is trend? A trend would be something like a data platform where we've seen it evolve over the last decade or so. Yeah. That's
fantastic. All right, so that's what you're gonna talk about tomorrow . And I know, what are you seeing at the conference?
What
am I seeing?
Yeah. First of all, it's grown.
Oh it's tremendous. I don't know how many vendors are here or how many he. Participants, but it's quite interesting. I've seen a lot of people come here to see what the various trends are, right? So they can differentiate, what's gonna be here in the long term versus what's like technology that may not be here for.
The near term and that's it also, there's quite a few private equity folks here, VC folks here as well. There are that I've run into,
and that's the thing that's I think, distinct about the health conferences. And health is obviously just partnership with health and Chime is they attract that community.
In fact, I was just talking to a venture capital person and we were just talking about we ended up talking about generative. Quite frankly, this is why we joked about having a punch slug bug kind of thing. Because I would say of the 12 interviews I've done so far, at least 10 of 'em have mentioned ChatGPT 4
Did they mention the problems that are likely to be solved or did they just talk about the. The hype around it.
Well, it's interesting because I think this crowd generally is talking about the potential, right? We understand that there could be some security issues, yet privacy issues, yet there could still be some hallucinations.
As they call 'em. Yeah. So we understand those limitations, but I think the thing that and you've played with it.
Yeah. actually use it. I'm actually a subscriber. Yeah. I have been for some time. So let me give you a real quick use case. For marketing material for Clearsense We needed to know how many beds per client.
So we have 26. I went to chat gtp. How many beds in the following organizations? Comment them out. Less than 20 seconds. I got that count. I then did the, what I was supposed to do, did the research, went to the websites. I got within 2%. So, either chat GTP is as smart as me or I'm as dumb as chat gtp.
We gotta figure that one out and. Okay. $3 to the organization, .
We don't have to really try too hard at this point. I think the reason people are excited about it, first of all, is they're not making the promises. So when IBM came, you and I have talked about this. Yeah.
When IBM came into the industry with Watson, it was loud and. And we now can use it as a case study of saying, don't come into healthcare saying we're gonna replace doctors. And yes, it is passing exams and it's continuing to advance in those kind of things. Let the world find out if it's gonna solve cancer or not.
Just introduce Yeah, the tool set to people and see what they can do. And I think the thing that's happening is my experience is the same, which is, okay, the hype reaches me. I go out there and use. And by the way, the hype reached me when it was G P T three and I was using it through a third party application that we were using it for our company.
And that tool was effective for us, but it was not great. We used to say use this tool for a crappy first draft. Sure. It's called Jasper ai. So Jasper AI sits on top of G P T three and I'm sure it'll be 3.5 or four here shortly. And it was a really good tool. It really helped us, it helped us do outlines.
We take these transcripts, put it in, and it does a lot of that stuff. And so we've been using it for a while and then all of a sudden, three five comes out and it felt like, it was like the overnight, like three, five comes out, and then four comes out. And the thing that's amazing to me is we went from crappy first draft to somebody was just telling me that they had professors look at eight papers, and four of 'em were written by students.
Four of 'em were written by gpt four, and the professors couldn't tell which ones were which, and we're getting to that level. Quality, which isn't that a pretty astronomical move in a very short period of time?
I think that it's the, the tech behind, it's always been there. Yes. The refinement for what we're seeing is evident.
Nothing in the history of mankind from a technology. And or from an application perspective has ever gone from zero to a hundred million users in two months. That's something to think about. So, and a few years back, when we look and, bill, I'm a professor as well. And I will say that from what you just described, if it's a short piece of work, that's correct.
But if it's a five to 10 page, no, that's not correct because, but it's, but
it's multiple prompts. You break it down. Yes. Let me ask you this from a programming standpoint. Did you program my program? I've written some code through it. In fact, I took a project that I paid a thousand dollars for about a year and a half ago.
And I thought, you know what? I'm gonna take that project. I went out to like a third party psych out developer to write some stuff for me, and I decided I'm gonna do the prompts and just walk through it. And I had to fine tune the prompts, but after I got the prompt, it generated the code that I paid a thousand bucks for
also, for debugging.
It's worked well for debugging as well.
What about documentation? Can you feed code through that maybe isn't documented real well and have it documented? You don't
know,
or I've not
done that one. Yeah, that would be interesting. It would seem to me like that would be something you could do without much risk because it's not gonna modify the code, it's just gonna put in the documents.
Well think about the use case that's being described, here where that interaction you have with the patient is a hundred. To recorded that information is then retrieved, and then yes, at that point, could we do something from a generative perspective or we actually generate a clinic note based on, that conversation happened between the clinician and patient.
I think you'll see some of that. So I think from a burnout perspective, if you will, or from a labor perspective, there's some good potential for that, but we'll have to see it beats a scribe. It would be a. Having to have a scribe, behind you doing, all this kind of business and whatnot.
Well, I'm wondering, so I mean, if I'm coaching all these startups around here, I'm saying, tell me about your business model. Tell me what you do. Is there a chance for utilizing these large language models and generative ai? And if there is, you should, be plugging into an API here pretty quickly.
You know what you need to do When we're done with this conversation and I have done it, you know how we go from booth to. We ask people what the company does and whatnot. Then you ask somebody else, it's completely different and you get like three or four different.
If you take some of these booths, go talk to those people. And then when you go later tonight, run it through chat g p t, you'll get a really accurate depiction of what they do probably better than the folks in the booth. I have done that and it is quite interesting. Yeah.
You're doing some interesting stuff with GPT four.
That's the thing that's amazing to me is it depends on their perspective and they come to me and say, this is how I used it. It's prior authorizations, it's, well
Bill, what if you have a set of data in, you're a researcher, I want to know how many patients in my dataset are over the age of, let's say 50 that are diabetic, type two that are on certain medications and so on and so forth.
If I want to count, what a better way to get that count, I can either, Or, type it in and get that response.
So trend or trendy.
Oh, this is definitely a trend that it really, honestly, if you talk about it from a call center perspective and everything else, it started with the ivr.
I do wanna talk to you about it from a call center perspective.
CCPT four has a lot of potential for a
call center, doesn't it?
Oh, absolutely. Especially for inform. And the really cool thing about it is because of what we've done from a language perspective many times on the receiving end, they have no idea that they're talking to a, artificial agent and whatnot.
And as we get better and better, that's even, becomes exponential. So I can use this to give information, let's say. And it's really interesting you asked that because I asked for, can you give me this information at a fifth grade? At an eighth grade level, at a graduate level, and I can get all of those levels as a generated response from that technology.
So,
can generate the script that a Oh,
absolutely. Health center agent. Not even that, but it can actually go back out. So you hey name that healthcare organization, my mother has been recently described with, dementia. Can you tell me a little bit about, dementia and what I can.
Absolutely. From an informational perspective, especially in a curated data source,
How much better is G p t four than like, say, Siri or Alexa? I mean, could it
Sure. So you're basically doing natural language understanding, but then you're doing the handoff so that you can actually, retrieve so Siri's kind of telling you something that's already there, as opposed to putting something and assimilating information to give you a picture.
So if you ask Siri to List a whole bunch of conditions, you're not likely to get, any type of diagnostic response unless they've already curated that from a backend perspective. It,
it feels to me like Microsoft just did what no one thought they could do, which is they got first mover advantage over Google in a space that Google was fairly competent in and now they're gonna be able to integrate the open AI tool set into all those different, I mean, Microsoft touches everything.
Yeah. All those different areas is Google going to be playing catch up on all this for quite some time now?
Well, I think, Google has always been, on the cutting edge and they were, they just didn't, put it out, soon enough, if you will, they talked about it a little bit.
But Google is really good for cutting a technology, but they don't necessarily take care of you when they switch to something else. You kind of get left, hold in the bag where know, on the Microsoft side, you're as versions come out, you're not left behind. If you will.
So, from a visualization perspective, look at Power BI on Microsoft side, look at Tableau and Click and others that you'll actually describe what you want that visualization to look like from a data perspective. And it'll be created for you, which that's, quite significant.
So I look at this from a data retrieval. Kind of my thing to be able to interrogate a curated data set and get that information back without having to, do the keyboard thing and spending inordinate amount of time. I see it as a time saver for a lot of the tasks that we do. Currently
it feels an awful lot like the start of Star Trek, when they would say, computer, do this, computer do that.
If you look at the progress we've made in the past year, I, somebody was saying to me, where do you think G PT four will be in 10 years? I'm like, can't even, I mean, literally, I cannot imagine. Cuz you take that model and there's no way for us to train it, by the way. I mean, that's not truth.
You can train it through prompts, but. It doesn't remem it has no long term memory on
the, on the prompts bill for healthcare. I like the technology of chat GTP from a interrogation perspective. But I would like that technology on curated data sets. Yeah. That we've created and curated.
I would love that. Cuz then I have, some assurance that what comes back is gonna be as accurate. Possible. And that's really important and I always talk about this bill. We as human beings can make mistakes of being forgiven. The technology gets one shot at it and boom, you got a little bit of a problem.
We'll get back to our show in just a minute. We have a great webinar coming up for you in April. We just finished our March. On April 6th at 1:00 PM Easter time, the first Thursday of every month, we're gonna have our leadership series. This one is on CSO priorities for 2023. Chief Information Security Officers, we have a great panel.
We have Eric Decker within our mountain, Shauna Hofer with St. Luke's Health System out of Boise, Idaho, and Vic Aurora with Hospital four Special Surgery. And we are gonna delve into what are the priorities for security? What are we seeing? What are the new threat? What is top of mind for this group? If you wanna be a part of these webinars and we would love to have you be a part of them, go ahead and sign up.
You can go to our website this week, health.com, top right hand corner, you'll see our webinar. And when you get to that page, go ahead and fill out your information. Don't forget to put a question in there. one of the things that we do, I think that is pretty distinct is we. like for today's webinar, we had 50 some odd questions that we utilized, in order to make sure that the conversation is the conversation that you want us to have with these executives.
So really appreciate you guys being a part of it and look forward to seeing you on that webinar. Now, back to the show.
Yeah, we've talked about that for a while. Yeah. Alright, so we've spent a lot of time on this.
Oh, I know. And you've done really well. We've got rough. $50,000 or something like that. . Yeah.
Well, we have raised a fair amount of money for Alex's Lemonade standing. Yeah. Captain's walking around here and everybody's getting their picture taken with 'em, which is great.
As we look at the challenges facing healthcare, there's a lot of discussion around obviously the financial challenges that health systems are facing. There's still burnout and clinician shortage. Challenges that are facing healthcare as well. And there's this overarching question, which is, how do we continue to improve the quality of care while being cost effective in the process?
What are you seeing or what are you hearing in terms of how we're going to be able to do just that, to improve the, continue to improve the quality of care? Cuz that's what we're. while doing it in a cost effective manner or even reducing cost out of the system.
Sure.
So, I always answer that with a data answer because we have all of the data that was required to make those decisions. Not just from a here and now, but also from a forecasting perspective. But we don't do it. We've been flying by the seat of our pants since the beginning and we're always looking in the rear view mirror.
We've got ourselves to a point now where we can't do that anymore. So by having a an environment of data that's actually, trusted and the requisite tools on top of that, then we can start making, certain decisions and whatnot. We gotta make those hard decisions.
There are verticals within our organizations that, that are. not Patient effective, nor are they profitable. We gotta make some hard choices and whatnot. From a documentation burden and so forth Florence Nightingale said it better than anybody document if you must, but only document that which is required.
We've gone nuts with documentation. We've, yeah, there's some regulatory requirements and there are some clinical needs. Let's not do any more than that. Let's not use our EMRs and, other systems as this crazy, data collection system for data elements that we're not using, we know what the critical data elements are, both financial, both operationally, research and clinically.
So let's just stick to that and, we'll be in a lot better shape. There's too much mess, if you will, that we gotta get down cuz it's a situational awareness. Exercise a little bit about the past, what's going on now, and what's likely to happen in the future, and then we gotta make decisions based on that.
Is there a way to look at the data? When we talk about clear science, we're talking about data transparency. Is there a way to look at the data and see, hey, we collected this data but we're not using it some of this is gonna be a moot point cuz there's regulatory requirements that require us to collect data.
Absolutely. That we're not using. Absolutely. But is there a way to identify the data that, hey, you know what, we've collected all this data and it never
shows up anywhere? It never shows up anywhere. So we did something like, we did something very similar with order sets. Remember when we started the emr?
Oh yeah. We built. Thousands of order sets and then we did inventory and we found out we're only using like, whatever that percentage was. It's like, and you kind of get rid of 'em. It's in the old days it was like
reports. Yeah. It's like absolutely. It's like, hey, we're up to 3,800 reports. It's like 3,800 reports.
What is everyone using those reports like and then you do the inventory and No,
what I used to do with those reports is I would write, Hey, in bold in the report, in the middle of it, Hey, if you're reading this, please call. Because I want to talk to you about this report and I'd never get any phone calls from most of these reports.
So I would delete 'em and nobody would ever know. So can we do that from a data perspective? Yeah, we absolutely can. What are we bringing in that we, we don't need? But more importantly, the organization from a governance perspective, what do they have that they don't even know that they have that is absolutely purposeful and actually can help them?
Hey, we've got clinics that are operating these hours. They need to change to these hours. We've got practices that are just minimal.
We I talked about that with Rob d mache from U P M C, former CFO for you, U P M C, and he's just talking about we. Yeah. Cause I was saying, do we need more data?
And he's like, he goes, probably not. We probably have enough of the data. We need to get it into a form and get it in front of the right decision makers so they can look at it and say, you know what? We put those locations in the wrong place. They should be here, and here.
And we would actually, just by moving those locations and ending those leases, we would increase access, we would increase our quality. We would increase the population health in
those communities. Yeah. And the simple discovery of best practices in the. We have evidence-based practice, we call it that.
Many times it's eminence-based practice. Who is the person that is noted for this type of therapeutic endeavor regime and whatnot, but, Kaiser and others look at their data and they find the best practices within the data, then they implement 'em, and of course the costs come down, the patients do better and so forth.
So yes, I think these data sets that we have will absolutely tell us what we need, but they gotta be put in a way that's accessible and more importantly is trusted so that when we get in a room and somebody sees something on the wall and they challenge. It's correct, not they challenge it and they, yeah, they're right.
And now we've just lost all credibility and we start all over again.
Well, Charles, it's always great to catch up with you and talk about G P T four and just to have that conversation about G PT four was really helpful to me and just understanding where G P T four is going in the future. And there's three more dollars for Alex.
There you go. Understand and I really appreciate. Thanks
man. Appreciate. Yeah.
📍 And that is the news. If I were a CIO today, I think what I would do is I'd have every team member listening to a show just like this one, and trying to have conversations with them after the show about what they've learned.
and what we can apply to our health system. If you wanna support this week Health, one of the ways you can do that is you can recommend our channels to a peer or to one of your staff members. We have two channels this week, health Newsroom, and this week Health Conference. You can check them out anywhere you listen to podcasts, which is a lot of places apple, Google, , overcast, Spotify, you name it, you could find it there. You could also find us on. And of course you could go to our website this week, health.com, and we want to thank our new state partners again, a lot of 'em, and we appreciate their participation in this show.
Cedar Sinai Accelerator Clear Sense CrowdStrike, digital Scientists, optimum Pure Storage. Sure. Test Tao, site Lumion and VMware who have 📍 invested in our mission to develop the next generation of health leaders. Thanks for listening. That's all for now.