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.
The need for the data to be suitable for building out those machine learning and AI models. It's gotta be clean, it's gotta be trusted. The deal with AI and machine learning in healthcare it can't be black boxed. It honestly, it really can't. The clinicians have to be part of the process from both the data acquisition, the data cleansing, feature selection, model build. And they've gotta be part of the process as you're building out the model. They've gotta understand and they need to be able to put some input. I'd rather have less false negatives and I can deal with some false positives.
It's Newsday. My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of This Week Health, a channel dedicated to keeping health IT staff current and engaged. Special thanks to CrowdStrike, Proofpoint, Clearsense, MEDITECH, Cedars-Sinai Accelerator, Talkdesk and DrFirst who are our Newsday show sponsors for investing in our mission to develop the next generation of health 📍 leaders.
All right. It's News Day and we are getting close to the end of the year, and we have Charles Boicey in the house with Clearsense. Charles, how, how's it going?
Hey, everything's good, bill. Good to be back with you.
I saw that you just spent some time in India with the team over there. How did that.
It went really well. we've got about 35 developers in, in Hyderabad, and I've been going back and forth to India for probably about 15 years now. So the difference in India, bill, is that we can put applications out there in healthcare there's very limited regulation and they're really eager to try new technologies out, and you don't have to go through the months of red tape that you do.
No, that's fantastic. And I saw you dancing on social media, which we've all been there, we've all been at those team building events where they say, okay, get up and dance. And we're thinking, well, maybe I, I just received a phone call. Maybe I have to step out, maybe . But no, you were a good sport. You hung in there. You did the dancing, and then. Of course your head of marketing put it out there for all the world to see.
Of course Costner did dances with wolves. I do dances with difficulty . That was about the best I could do. Yeah,
man. It is. It is what it is. So we were gonna try to do, and well actually we'll start with this topic. We were gonna try to do this as our first. Interview for this week, health in the Virtual World. We're actually gonna do a, an interview where you and I both went into vr into the metaverse. We met up there and we were gonna record our first interview there. I, quite frankly, I'm just too busy to set aside the time to get everything all set up.
but the next time around you and I are gonna do that. That's a. Precursor or a little teaser for what we're gonna do in the future. But let's talk about it. So you and I are early adopters. Your closet is probably full, like mine is of 17 different watches and different sensors that you've tried out and whatnot. But now we both have headsets. I've got this one right here, right on my desk, and I'm searching for what the business application and what the healthcare application's going to be. What's your first take? I mean, you've, you've been in this probably a little longer than.
Probably the first thing is that in the virtual world, I look so much more like Brad Pitt than I do in the real world.
Cause you can design your avatar, however you wanna design your avatar.
It's pretty, pretty amazing. So yeah, there's a whole different world out there. It's not necessarily an escape, although it is for some, I think there will be some practical uses for it at some point in time. I think from a. Training perspective. I think there's a lot to be said for using that environment from a training perspective.
Yeah. The killer app right now is education.
Yeah. and I'll tell you, during lockdown in India all education was provided by Conversational agents bought technology as well as vr and it was extremely effective. Very, very effective. In addition to training as we spoke meetings are pretty cool in a VR world in a virtual world as well that's not quite as developed, but as it gets more and more developed also if you think about. You remember back a few years ago, Google came out with these interesting glasses and whatnot,
Yes. That. Really showed us that this is going to be a trend going forward, not necessarily trendy. And I think what you just showed is a little bit too clunky. Clunky to be relying on it 24 7. I think there needs some, be some more work done there. But I think we're getting pretty close to it.
Yeah, it's, it's a couple things there. One is, it's really interesting to me, Whenever you look at technology, what you have to imagine is unlimited bandwidth. You have to imagine unlimited miniaturization and unlimited processing power and, and cost bill. Yeah. And cost gets driven out, right? So right now you can get a a headset for about 400 bucks.
But you know, they're climbing a little bit in terms of things they can do, if you drive all those things in, then it's not this big old headset that's sitting on your head. it is a pair of glasses that has augmented reality or virtual reality depending on what you're trying to do.
I got one for my birthday and I was trying to figure out what the business application was. What's the application where I go out and buy one for the 10 people that are on my staff? And I didn't really come up with it yet. This is still really early on, but what I was looking for is, When meetings become better in virtual reality than they are via Zoom, and we're not there yet.
But the thing where it does Excel is education, right? So, the example I keep giving, I give two examples. One is, Hey, today open your books. We're gonna study king Tut's tomb, or open your books. We're gonna study the Holocaust as opposed to put on your VR headset. We're gonna walk through King Tut's tomb, or we're gonna walk through Auschwitz.
I mean, they have completely different engagement scores, completely different retention scores. I mean, if you walk through Auschwitz, I guarantee you you're gonna. Your study of the Holocaust as opposed to reading something in a book?
Absolutely. Absolutely. And what they found was the systems in India that be utilized, were able to give extremely Granular reports on the individual students and how well they were doing even much more so than the teachers in the classroom. So it was it was a really good experience for them.
we talked about some of the limitations. The, it's a little clunky now. It's a little big. The software's still early stage. In a lot of cases it's still interesting. I mean, whenever I have somebody over to the house and I put it on and they just look around and they're like, oh my gosh this is this is unlike anything I've ever seen before.
So it's not like the software is pixelated or anything to that effect. It's, I mean, you could stand on the top of Mount Everest and do a, do a 360. It's pretty, Pretty amazing that way, but the, the software interaction is still a little bit. Clunky.
Yeah. And Bill I've been doing this for almost 10 years now. And even in the early days we would simulate, we would, of course we'd have everybody wired up and monitor 'em from a physiological perspective. And one of my favorite simulations was were bringing somebody up in a balloon. And then dropping the bottom of the basket and simulating a free fall and then just monitoring their their heart rate respirations.
That's just mean. Yeah.
So a lot of that early stuff was, was a lot of fun. Or we would place horizontally and then simulate them flying through the Milky Way and whatnot, which was kind of interest.
Your body does react. It does as if it's actually happening. And that's one of, that's one of the applications to healthcare. Obviously we talked about education. That is one of the primary applications to healthcare both on the HR side, on the onboarding side for nurses on physician training in the future and whatnot. That's the killer app right now. But also in the care of patients. We saw this at Cedar. Where there's a couple of startups that are using VR to patients that are bedridden, given them the opportunity to go outside and your body physiological response is, I was just outside. It's really kind of interesting.
It is. It absolutely is.
So a lot of, a lot of healthcare application as we move forward. And you and I will keep a, a close tabs on that, but hopefully coming out of the new Year, I'll have some time to play around and, and we'll get the next interview. You said a phrase here, trend versus trendy. It's the first time I heard you use that phrase. What spurs you to start talking about trend versus trend?
We see a lot of trendy things in healthcare that kind of go away, especially on h i t and I'll have to give one of my colleagues Amy Webb the credit for this terminology. And she's a futurist. She has an institute and everybody can look her up.trendy. Going back as far as:
If we're gonna get wide adoption absolutely not. So I always use that and she uses that as an example of something that's trendy, something that's a trend. We saw this very early on with initially with Uber Airbnb, where you now have the control. There's no longer any travel agencies, right? There's no longer you don't have to stand on the street corner and hail a taxi. Now even the taxi services have adopted this Uber type experience and whatnot. We're even seeing that in healthcare, right? I think that is an absolute trend that will will continue. Will the VR thing be a trend or something trendy? I'm gonna say it's gonna be a slow process from a, a trend perspective.
Yeah, I'm, I'm more bullish, I think in the next 10 years, well, for the people listening to this, in the next five years, they will all have put on a headset and be doing something in virtual reality.
But I think the general population, I'm telling them within the next decade they're all gonna be operating in, in virtual reality. And they look at me like, why would I do that? I'm like, I don't know. Why do you do Zoom? . Yeah. And they just look at me like, well it's like, oh, that's just the norm. I'm like, yeah, that's how you'll feel a decade from now.
Can you imagine a patient from an elective surgery perspective being brought through what to expect when they arrive? Through the the whole pre and post-op. All of that can be done from a future reality perspective as opposed to sitting just like you and I are talking back and forth or just showing slides and whatnot, they'll know exactly what to expect and what's expected of them.
Yeah. We will look, in a decade we will look at this experience, you and I talking via Zoom and we'll think, oh my gosh, how archaic. And that's, that's my feeling on vr cuz when you meet somebody in, in vr. It feels like you are in a room with them. You're sitting at a conference table with them.
You're having a conversation now. It's still avatars today. I think that will change somewhat as well. So we'll, we'll see where that goes. Trend versus trendy. You brought up car stuff, so autonomous driving, I think somebody would ask is that trend versus trendy? And we're seeing enough technology out there that it looks like it could be a trend.
But the problem is if a car kills one person, they're gonna shut it down as opposed to the millions of people that die every year from another person driving into them. So is that trend or trendy?
So here's the, here's my deal on autonomous cars. It has to happen all at once. You can't have a bunch of autonomous cars interacting with humans just the humans.
They can't calculate. Humans are gonna mess it up.
Yeah. Humans are gonna screw it all up. And a motorcycle will really mess things up. So I think it's a, a trend. Will it happen? I think we're quite a, quite a, a long ways off. Cuz we do have autonomous vehicles out there. But it's still in the experimental stage, so we, we will.
Yeah, I, I think it's, I think it's trend and I think what you'll see it, you'll see it not used in city driving. You'll see it used in long distance driving and that kinda stuff. And somebody was telling me they, they used their auto, it's not autonomous, whatever they call it, but autonomous driving for, for our purposes to go from one place to another.
And they said it was a lot slower. I'm like, well, why is it slower? It's like, well, you know how we like pull up to a stop sign and hit the. Break and we stop, well, autonomous driving sort of inches up to it slows down a lot earlier. It inches up to that spot. Because it has to be more cautious of its surroundings.
And because the program can't be as aggressive as we are it's actually a slower experience. Now, if I'm going from, I don't know, if I'm going from Kansas City to Denver you just, you put it on, you go to sleep, you wake up, you're in. That's hopefully, yeah, hopefully trend versus trendy in healthcare. Gosh, this is a, this is an interesting one because there's, there's so many trends going on, but there's so many things driving those trends.
📍 📍 All right. We'll get back to our show in just a minute. We have a webinar coming up on December 7th, and I'm looking forward to that webinar. It is on how to modernize the data platform within healthcare, the modern data platform within healthcare. And I'm really looking forward to the conversation. We just recorded five pre episodes for that. And so they're gonna air on Tuesday and Thursdays leading up to the episode. And we have great conversation about the different aspects, different use cases around the modern data platform and how agility becomes so key and data quality and all those things. So great conversation. Looking forward to that. Wednesday, December 7th at one o'clock. Love to have you join us. We're gonna have health system leaders from Memorial Care and others. CDW is going to have some of their experts on this show as well. So check that out. You can go to our website thisweekhealth.com, top right hand corner. You'll see the upcoming webinars. Love to have you be a part of it. If you have a question coming into it, one of the things we do is we collect the questions in the signup form because we want to make sure that we incorporate that into the discussion. So hope to see you there. Now, back to the show.
📍 📍 hospital at home trend or trendy?
Oh, that's a trend. our colleagues in the UK are really going forward with this know, we'll, we'll follow along and we do have organizations that are, are working on this. in the UK it was primarily done for, and it's being done for. Purposes of nurses and others not having to work in the hospital. They'd rather work in somebody's home. It's much more advantageous. So I think you need to see some correlations there. The large device manufacturers, Phillips and others are definitely gearing up their product lines for that.
And as we get better and better with virtual reality, there's gonna be connection there where folks are gonna be able to be monitored and they'll be higher acuities acuity within the home. But I think I as a nurse would much rather do my shift in somebody's home than in a hospital.
Yeah, that's a good one. Clinical AI trend or trendy
it's definitely a trend and our colleague Dale Sanders put a piece in recently in LinkedIn that worthy of everybody looking at and really discusses the, need for the data to be suitable. For building out those machine learning and, and AI models. It's gotta be clean, it's gotta be trusted. So I think that is really important. The deal with AI and machine learning in healthcare it can't be black boxed. It, it honestly, it really can't. The clinicians have to be part of the process from both the the data acquisition, the data cleansing, if you will feature selection, model build. And they've gotta be part of the the process as you're building out the model. They've gotta understand and they need to be able to put some input. I'd rather have less false negatives, and I can deal with some false positives I think that's really important that this stuff is approachable and explainable and the real downfall with AI and ML in clinical practice is we as humans can make mistakes, right? And we're forgiven. The machine gets one shot at it. and I think the other thing that is really important these models, one, have to be tuned to the demographic.
What works in Florida, where we are is probably gonna need to be tuned in New York City because the, the demographic is much different and you can't just deploy these things and walk away from 'em. They actually have to be monitored, tuned as time progresses.
Charles, I was struggling. I'm trying to come up, come up with something that I think is trendy in healthcare. Do you have an example of something that's trend.
what is trendy and what we're moving towards is understanding that the future for us is a state of wellness. We're getting there slowly. And what I mean by that bill, right now, we don't provide healthcare, right? We provide, we provide sick, sick care. Yeah. We provide sick care. But as we continue to wire people up, take care of 'em in the home and so forth, and be able to
But that's a, that's a trend as well. Everything I came up with, I'm like, that's a trend. That's a trend. That's a trend. This is a little controversial, but and I'll, I'll throw it out there. Health equities feels to me to be trendy, like everybody's jumping. But, and I'm not saying that we shouldn't be doing this. I, I think we should, but I don't think it's backed by, first of all, a conviction from leadership.
I don't think it's backed by compensation models to leadership. I don't think it's, so that's one of those that I think that has the ability to be trendy right now because it doesn't have the underlying conviction and drive behind it that it needs to, it's. Trendy to talk about and everybody wants to get on stage and talk about it and try it out.
Hey, here's what we're thinking. But you know, in order to follow through on that, I think it needs that underlying foundation of people with strong convictions who are pushing it. Were compensated to do it and whatnot. And again, I know it's controversial to say that's trendy. I'm not saying it's trendy because it shouldn't happen. I'm saying it's trendy because it doesn't happen. That underlying push.
Yeah. You see folks going into it and then retreating out of it. Right. They're talking about it, but you know how much How much, yeah.
As soon as financial times get tough, it's like, well, we're gonna go back to really focusing in on making money.
Yep. Yep. No, I, I totally agree with that.
Yeah. So that's a, I mean, that's a tough one. 📍 📍
Conference season is upon us and our this week, health team and I will be at the Chime Fall forum celebrating their 30th year in San Antonio. And we're also gonna be at the HLTH conference, HLTH in Las Vegas the following week. While at these events, we're gonna be recording our favorite show on the road, which is interviews in action. And as you know, what we do is we grab leaders from health systems, healthcare leaders from across the country. And we capture 10 to 15 minute conversations with them to hear what's going on, what they're excited about, what are their priorities, and those kinds of things. It's a great way for you to catch up very quickly on what other health systems are thinking and doing across the industry. We actually air this on the community channel this week, Health Community. It's the green one. So if you go out onto your podcast listener of choice and do a search. This channel is also where community members like yourselves have been invited to do interviews of their peers. So check those out as well. You can subscribe wherever you listen to podcasts. Look forward to catching you on our interviews and action. 📍 📍 I'm trying to think what else is trendy or what has been trendy. The other one I was sort of toying with throwing out there was interoperability. It's so cr it's, it's not that it doesn't need to happen. Don't hear me say that. But trendy in that we've been talking about it for two decades. Yeah. It's like if it was a trend we'd, we'd see a significant amount of progress.
Yeah. And we're not, we're not, we haven't, although we've created some new standards though, haven't we? We have. Yes, we have. And we talk about those new standards. But is there widespread adoption where they're actually interoperating with each other? No.
Right. Charles, I'm gonna go to this post. Dale Sanders. So why is this Dale Sanders chief Strategy Officer at imo Intelligent Medical Objects. That's right. Yeah. So, yeah. So why is AI ML underperforming in healthcare? We need a national strategic emphasis to improve the quality of data we're collecting in EHRs including clinical notes and longer term strategy to collect more data about patients beyond the ehr.
EHR data quality is a poor representation of individual patients as well as the population. EHR data quality is not good enough for AI ML models to be patient specific. It is good enough to be directionally informative. AI ML can still be valuable to EHR data if we apply.
It to accelerate the human situational awareness and hypothesis generation process. That is to accelerate the what's going on here. Process. To use a metaphor, we should stop trying to use the HR data with AI ML to drive the car and instead be content for now, but with simply informing the driver. I'm gonna stop there for now. Let's talk about EHR data quality and its use for AI and ml. I assume you agree with this a hundred percent.
Yeah, I do. From a data quality perspective, many organizations haven't done the due diligence from a ontology perspective. And again, it goes back to that building out a model in one organization and transforming it to another if the proper due diligence hasn't occurred from a data quality perspective.
Not so good. The other thing you talked about Bill, is you and I can be diabetics in a category of type two diabetes. But each of us has a completely different profile as far as where we live, what we eat our exercise, all of that. So getting back to an 1 requires more than the data that's resides within the electronic health record.
It requires data that we're generating 24 7. The other thing that he brought up that's of equal value is the clinic notes. All the unstructured data that has to be taken into consideration as well to build out effective models. And I think lastly this whole term of prescriptive is a problem. Those that prescribe do not wanna be prescribed to. But if your models will tell me enough, give me enough information, it helps me make a decision, then that's welcomed. But you know definitive type algorithms, no. No.
Charles, you and I have talked about the data quality and the use of il and I've also had conversations with John Hopkins and others. The telemetry data, the imaging data, that seems to be the data that is of enough quality to do AI and ML models on
It's not enough. Right? Right. Yeah, it isn't enough.
Yeah. It doesn't, it doesn't provide context. It doesn't, all that stuff and all that stuff's in the ehr. But that's when you look at some of the models that clinical AI we're talking about here, I mean, you could do administrative AI and Oh, sure. Of course. till the cows come home? I don't know. But you could do that every day of the week, but the clinical models have to be accurate. And what I'm seeing is taking the ecg taking the image and whatnot and literally running those through models that are. As accurate, if not more accurate in some cases, then the human who's reading like x-rays was an example that was given to me recently where they're applying models to it. Because first of all, Physicians are, they just don't wanna read xray anymore. So it's actually a dying kind of skill set.
And the machines not only read, but they can identify things that humans generally are overlooking or overseeing. And you can actually lay those things next to each other and say, human read. Machine read and you could look at quality over time.
I would rather see that as people consider that as, especially in images, as adjunctive not necessarily an outright read, but helping pathologists and or radiologists pinpoint areas within those images that they should probably spend a little bit more attention to as a definitive, Hey, it's this, that, or the other thing,
yeah, it is, it's augmented intelligence, right? So we're augmenting the intelligence of the clinician and helping them. We're removing some of the cognitive load. These are people who are looking at patient, patient, patient, patient. Their workday is full of context switching, right? They have to go from this room to this room, to this patient, to this patient.
Research would tell us that the cognitive load of all that context switching is mistakes. Mistakes, oversight, tiredness, that kinda stuff. And so if the machine can look at it and say, Hey, here's the three things you should focus in on, and they look at the screen and go. Oh yeah, that is, that that is a problem.
Right? Then they can utilize it. That's how I think we're gonna see it implemented. The question becomes, how do we come up with more quality models? And we need to get better data. We need to be able to validate the models early and often with as much data as we possibly can. And then, as you said, The models are very specific. They're, they're geographically specific. we talked to Michael Pfeffer about this. He's like the ED from 10 o'clock at night to eight o'clock in the morning is a different population than the. Ed from eight o'clock in the morning to 10 o'clock at night.
And so we have to, we have to normalize for those kinds of things. And then we have to understand that every patient is a distinct set of genomic sequences and whatnot. And so you have to get we still have a long way to go here. I mean, yes, we do. To really tap into this, there's, there's still a lot of work to do.
What's the one or two foundational items you'd like to see tackled in let's And, say the next five years in order to really set us up for a future where we're tapping into AI?
I think number one we get to the point where we are actually collecting data 24 7 from all of us. I think that's absolutely essential for higher efficacy.
So sleep data, food data, walking, all of it.
All of that. And I think the understanding that these algorithms, these models are gonna have to be basically open source. The idea that this is IP, I think we have to get somewhere beyond that. We just have just can't say this is, I'm sorry, I'd love to show you, but this is my ip, this black boxing of these models isn't doing anyone any good.
And we've seen all kinds of failure because of that. So again, from an IP perspective, the way it's deliver ed, the way that the information's pushed out. Absolutely. But the actual models and whatnot really need to be just like HL7, just like FHIR standards. Something that can be shared. I know that's kind of heresy in the,
In some circles it would be. Yeah. Cuz there's, there's a free market economy,
and bill, this is all math. Math is not patentable. You can call this machine learning, AI but in the late eighties, early nineties when I did this, we just called it math.
One of the stories we were gonna cover was Salesforce launches Patient 360 for health, and we're not gonna cover it, but when you talk about collecting all this data, pulling all this data together, that Patient 360 talks about not only the the clinical profile of the person, but also the demographic profile of the person and some other stuff. Where is all this data going to reside? We're not. I, I can't imagine we're gonna start putting all this sensor data and demographic data and purchasing data and into the ehr. Where's it gonna go?
This is your healthcare data platform. This is synergistic to your existing edw, synergistic to your, to your emr. This is the the data lake technology the streaming data technologies. This is really this third space if you will I've called it for years, the healthcare data ecosystem. So this is really. What it is. I don't see epic or Cerner being able to develop this out in enough time.
So anytime soon. Yeah, and Salesforce is launching, I guess the precursor for housing all this data as well. So we'll see what happens. Charles Day before Thanksgiving we're recording this. Thanks for taking the time. I appreciate it and I hope you have a great than.. You as well look forward to:
Absolutely. What a great discussion. If you know someone that might benefit from our channel, from these kinds of discussions, please forward them a note, perhaps your team, your staff. I know if I were a CIO today, I would have every one of my team members listening to show just like this one. It's conference level value every week. They can subscribe on our website thisweekhealth.com. They can also subscribe wherever they listen to podcasts. Apple, Google, Overcast. You get the picture. We are everywhere. Go ahead. Subscribe today. We want to thank our news day sponsors who are investing in our mission to develop the next generation of health leaders. Those are CrowdStrike, Proofpoint, 📍 Clearsense, MEDITECH, Cedars-Sinai Accelerator, Talkdesk and DrFirst. Thanks for listening. That's all for now.