81 – Tarun Kapoor on AI’s Role in Modern Healthcare

0:00

if you enjoy this content please take a moment to rate and review it your feedback will greatly impact our ability to reach more people thank you all right we are thrilled to welcome back our friend TK to ruin Kapoor from virtual health to ruin how are you brother are you Rick Marcus glad to see you both I know you’ve been getting a couple of vacations in every once in a while well dessert but I was I was telling him as we were chatting before your your podcast is predigious I have trouble keeping up because there’s so much going on and but you do such a great job of keeping me uprised of what’s going on so thank you and thank you for the opportunity to come back we love having you back Vic is taking some good notes on the show so I’m gonna let him lead and I’ll chime in with color but let’s jump in we have a ton to talk about yeah yeah well you know you’re one of our returning guests and so really our expert in the intersection of health systems in the clinic actually how can we use these tools AI tools to benefit patients beneficiary benefit our clinicians we talked previously about Virtual Health and its partnership with robot so maybe uh let’s start off with give us an updated how Virtua is doing and how your experience with the what partnership is going and then we’ll I’m sure we’ll expand into lots of other things around AI so overall Virtua probably just like every other health system what some result no such thing anymore it’s just busy all the time and I think we did talk about last time that this is a preview of the next 10 years at least from my cute care perspective there’s just going to be a lot of a cute care that’s gonna get just a lot of illness to have to work on and care for and now that we have even more sophisticated treatments than ever before there’s gonna be no lack of work for at least the next 10+ years and and the question now is how are we how are we gonna be able to provide services and and I you know I not only just an acute care perspective but you know Marcus and you I connected with you offline about some of the experiences you shared from a personal perspective and I wanted to say thank you because there is an absolute human connection here and an awakening that we’re all going through together and so we’re gonna have to figure out how to use these tools because it’s not gonna be easy easy to deliver uh in a legacy way there’s just not an option so we’re gonna have to do this differently so yeah uh things are busy here in Birtua before we dive into that I mean I wanna kinda double click on what you meant by the the volume and acuity is where at first I thought it was a pent up demand after the pandemic and it seems like it’s more steady state now so we’re gonna have this higher volume and higher acuity is that your view for the next several years yes and and if you look at the demographics experts out there they gonna tell you he told you so you can’t escape the numbers on this one so that the current projections based off of the curves are only the first half of the baby boomers of aged into the 65+ range only the first half so you got the second half yet to come in the moment you hit 65 and on your utilization of healthcare goes up typically around 300% and now you’re gonna be covering that also back doing that with a relative 17% decrease in working staff so the mismatch is not going to go anywhere for another 10 to 15 years and then as you get Frank Kavey Safavi who is educating on all this is and then the problem resolves itself from a global perspective for reasons that probably not we’re talking about in public but you can read between the lines and so therefore if we’re just gonna say well we’re just gonna train up a bunch of people today or we start the pipeline today let’s say from a position perspective you put someone in the med school today minimum bare minimum seven years before they’re able to do anything in an independent fashion and now we are could actually create some other problem of like a bunch of people we just trained up and there isn’t the same amount of work for them yeah the the the that volume that big Bolis of population need if you don’t if it takes eight years seven eight 10 years to train someone they almost miss most of it and then they come out as we are starting to have the decline so that’s what you’re saying right yeah so so like in there’s lots of fear around AI replacing jobs in the overall economy in healthcare it’s really interesting we we need the help when we we’re not gonna lose job people won’t lose jobs if they want to keep working and engaging we have plenty of work to do we need to help these people do more be more effective deliver better care and I think your partnership with robot was really one of the leading one of the early partnerships exploring that how can that work in behavioral health yes and we’ve Learned a ton so far and interestingly since we last spoke nobody even got a place on 60 Minutes and there was a feature on 60 Minutes about robot virtual got a clip so we had 30 seconds of fame and and you know it’s always good news when you’re on 60 Minutes it’s not because you did something bad is yeah that’s that yeah that’s that’s right that’s right now I didn’t make it on I didn’t make it onto the video clip so as a you know kid growing up with Indian parents that was a major disappointment because you know that is the epitome of being an immigrant an Indian immigrant kid in the United States is making it on to 60 Minutes for something good but virtue made it and robot did it and I think you know I give a lot of credit to the way 60 Minutes folks did it they did a really nice job of explaining the difference between open and closed models and the pros and the cons of them as so as we talked about on the last time I was here we started using robot late last 2023 which is a closed model assistant people uh not a large open model large language model um we started to offer to patients who had mild to moderate depression and or anxiety and the idea of our our hypothesis was if we could partner with a digital therapeutic company and do it in partnership with them and then prove results then can we go get a code a reimbursement code versus and and put the money you know put our money or the or mouth or money is or money or mouth is how it saying goes and show the results and we actually have some results that were now getting ready to release and I’m actually glad to preview some of them here with you all let’s go great excellent I like that alright so so yeah so um recently hot off the press again this I thought you know we would be at 500 active users and maybe I’ll get there eventually but we’re not quite there yet but we Learned a couple things along the way one of the things we Learned along the way is we’re not as good as we should be at collecting mental health data um in in in healthcare and and you think about it there’s just so much going on in a clinical visit that I thought well okay first thing we’ll do is we’ll pull all of our PHQ scores the PHQ Personal Health Questionnaire score which is a depression score and then the gad score tad stands for general anxiety disorder so if you hear me say PHQ is depression scale jet gad is anxiety scale we didn’t do as good as job as I thought we were doing and pulling that it’s no fault of our doctors there’s just so much going on in a 15 minute visit and honest reality is I don’t think this unique the Virtua you start asking those questions it starts to open a floodgate and that 15 minute visit is no longer a 15 minute visit and you’re already behind and so we got to figure out other ways of getting this information because it’s there’s unfortunately a time incentive not to talk about it and that’s just the reality you know just being candid uh it’s just the reality of the world do you do you think another aspect is that prior to this partnership with Robot you would collect the doctor would collect that information but there wasn’t really a place to refer a patient to that that didn’t have a long wait list and so you know it’s gonna take a long time lock off my schedule and then I find out that Vic has mild to moderate depression but I can’t help and and maybe it’s better not to even open that up is that fair or maybe I’m no I think that’s absolutely fair and I also think if you’re in a clinician’s office you know top therapy is the basis of treatment starting treatment top therapy doesn’t happen in the last five minutes of a 15 minute encounter alright you know talk therapy is extended and so sometimes the easy button is to write the script that you can do you can you know be prescribe something in 30 seconds with the main state of treatment for mild to moderate depression you can be assisted by medications but the healing happens with the cognitive behavioral therapy and and that’s what we’ve Learned so what we’ve seen so far of you know first about 200 users have used it generating about 14 sessions 10,000 minutes of usage we have seen for those who completed the course so far with at least 60 days plus for the PHQ which is the depression scale we’ve seen about a 2 point drop in their depression score which meets clinical significance and the P score for these this goal uh uh my fellow statistical geese out there is point zero zero three so it’s statistically significant legit significance there and then for the gad also not quite a two point drop just a little bit less than a two point drop but also significant both clinically and statistically and a couple other things that we see in here the usage uh is really interesting 76% of the usage has been outside of our clinic hours vast majority of it happening between 6 p m and midnight but we have utilization 2 a m 3 a m 4 a m so that was our hypothesis going in one of the things that was not in our hypothesis the amount of usage before 6 a m and 8 a m as people waking up getting ready to go to work I didn’t think that was gonna be the case I knew you may have trouble falling asleep but the thing that you waking up and your utilization um free work is almost the same as during working hours from a percentage uh in those dawn hours was really interesting to me but I you know I think what we’re finding is the tool can have an impact talking can have an impact and and and I think that’s what we find so exciting about this um I so first of all that’s fantastic um feedback that you’re that you’re getting that you’re statistically moving the needle a a couple just like brass tax questions um as someone who does therapy uh I never do the pH Q9 in session I always do it on an app outside um for the reason that you that you mentioned just because like there’s not enough time you know we gotta get to the talking right in our actual session um but the gad I’m not sure I’ve ever taken that um can can you just talk about what like literally what are the questions in the gad cause I’m I’m just not familiar with that one I think most people know the PHQ but what what what’s actually in the gad it’s exactly around the question that you’re mentioning Marcus around questions around like I feel stressed or I’m starting to anticipate I anticipatory stress about starting something oh okay and how um how frequently do you how frequently do you get that and then how what is the impact and does it start to impact whether or not you actually participate in something so almost from a social anxiety perspective I can bring you know I pull up the specific questions easily or anyone can but that’s what it is it and we know that depression and anxiety are very closely linked especially anxiety can cause depression right right in in in that link is so the interconnectivity of these things just don’t bucket easily say well I have one thing for this and one thing for that these things interconnect okay that that that was that was great and then just on the note of you you realizing you don’t do a great job of collecting this data but now rolling out this this pilot with robot does does robot fill that gap for you you know like are are are as you’re as you’re recognizing this deficiency and of course we’re talking about constraints of workforce you know a big part of this is how do we optimize the workforce and triage you know the most severe cases of of you know depression anxiety and other other related mental illnesses um to the people who need to be spending a lot of time with people and we use technology for the rest I mean is this robot app going to be you know part of your improvement in data in clinical data collection so us this is a very important what we call pro patient related outcome or sorry patient reported outcome and it’s only gonna become more and more um actually mandated just in July 1st or CMS mandated that if you’re doing joint replacements you have to collect periopera preoperative patient reported outcomes and then you’re also gonna have to collect at the 12 month Mark how to do that so this is this is going to become the standard of incorporating patient reported data into your planning if you’re only saying well the only time I’m going to ask is when they’re physically in front of me we’re gonna be held accountable for a completely different way of doing this which in a value based contracting world has been around for a long time now in a even a fee for service environment it’s starting to converge I would say so yes and and we’re just gonna and I think you’re gonna have to get people more flexibility to share their outcomes in methconescent outside of their patient portal that’s right right patient portals great but if I can widen the funnel by using another tool I need to get the data in I mean it’s really impressive results and and maybe as importantly you’re helping 200 real people in your in your treatment area get better and and feel better and have more productive life so that’s all really impressive and incredible I’m interested for you to compare if you can these results compared to the traditional in person talk therapy CBT therapy is it a similar size and scope is it hard to compare is that not even the right question how do you think about this um protocol versus the what traditional protocol I’m not sure if we’re ready to compare it uh to traditional because it’s again we’re we’re really still very pushing very hard on on adjunctive this is meant to be supplementary to this these are folks who were still encouraging to go see a licensed professional or psycho psychological professional where I do think and where we had to do some additional self analysis and this is where the data gets hard is what if he didn’t go did he still get better because a lot of people we may recommend to go and they don’t go either because there’s a stigma is usually the biggest one but let’s be realistic cost is not insignificant for psychological therapy many folks who provide it are out of pocket these days and then if you even do have somebody getting in so where we have to do the greater analysis and this is where I think partnering with insurance companies cause that’s a nice nice extent pieces well they’ll know if they’re reimbursing for it I don’t know um unless they only stay within my network but we may not have all the providers so I think that’s the next place representing the data back to say is your total cost of care going down on these patients because it is now we find up beautifully yeah yeah okay I think that’s fair it is I think an adjunct therapy is what we need I was trying to lead to kind of making the case that this should be reimbursed by someone and the frame of reference I have for that is existing therapies that whether they’re self pay or reimburse that there’s a there’s a structure there what I’m especially excited about for this and and maybe this is where I’m negotiating myself or or you know negotiating against anyone who’s in digital therapeutic space what I think is actually exciting about a digital therapeutic compared to a traditional pharmaclogic therapeutic is that the quote payer has to pay for the for let’s say a pharmaclogic therapy once the medicine is dispensed they have to pay for it regardless of whether the patient takes it or not right we can actually build out a model here to say I can dispense this thing because I’m just giving you an access code but reimburse me when they use it right it’s all right could could this potentially be without having to go full blown capitation can you use this as a fee for mom almost as hybrid of a fee for service model if you’re in a fee for service environment that then says what but only pay me if they’re using it and they’re not using it stop and if they stop using it stop I I I that if we can pull that off that I think becomes the best of all worlds here because now you’re paying for effective therapy not just dispensing a Bill that’s sitting in someone’s medicine cabinet yeah yeah okay I think that makes sense then the kind of the other side is um there’s a safety and with behavior health there’s a safety or um triage concern that that I know Virtua is was interested in well but I think it’s interested in did you see any signs of that and could you get people help when they needed to be kind of escalated or maybe 200 wasn’t enough but that’s a really important piece to just to follow at least I think it’s an important piece to follow great question and absolutely is one of the biggest issues that we had in our heads before we got started on it and so the good news is very very good news is that we have not seen anyone trigger language that required immediate level of attention but the system from our testing the system is able to pick up concerning language so it continues to help to produce that and for those who are still a little bit hesitant what I would ask is okay well if a patient calls and leaves a voice message to talk to the clinician and does your voicemail automatically detect concerning behavior it work right reality is medical advice messages through your portal a lot of those don’t get checked for 24 hours at least this is this language model was sorry this this um closed model is after able to detect some of that language immediately so far we are not seeing that yeah so you are um looking for critical phrases or words around uh suicidal thoughts or something something related to that and we don’t need to give any exact words but it’s screening for that and and it’s great news that the small number you didn’t you didn’t see any of those but that’s in that that security measure that safety measure is in place um to the best it can be in place with robot it is it is there and we so we feel very comfortable about that I mean the other thing is that we’ve also been selective in who we’ve offered this to we’ve focused on mild and moderate depression right if if a clinician when they’re recommending this product detects severe depression with thumb you know conservatives self thoughts harm thoughts we are staying away from it now that stated you can make an argument well then what are you still offering you don’t have anything could anything be beneficial than nothing I I I don’t know how to answer that one at this point um but yes we have decided to stay away from that for now but I I I think it’s it’s a valid conversation given the the simple fact that there’s such limited resources out there yeah it’s an unfair comparison and I wanted to ask it cause I think a lot of listeners will be saying it to themselves my belief is that getting people some help and help at different parts of the day at 9 at night or at 5 in the morning or whenever they need it is clearly in my mind a net positive and yeah I think AI tools and support tools are held to a you know ridiculously high standard that traditional um services are not held to as you as you say that if someone calls in and they leave a voice message that that that’s not checked for a little while but I wanted to ask it just cause I think a lot of people will end up digging into it so well you know there’s a lot of places we can go with this conversation and maybe I’ll hold this gone for later but I agree with you there’s this thought I think it’s a false the false promise that well it needs to be 100% accurate or 99.5% accurate people will adopt in healthcare it needs to be safe safe and accurate are two different things and we are very comfortable that this digital therapy we’re using is safe is 100% accurate no but the other question and one of our board members questioned me on this is well what is good enough in terms of correct and the question I pose back well do you know what the passing exams for is for the United States board exam and the answer is you need to get about 70% correct in order to become a board licensed physician so I think the number is probably should be higher than 70 but we don’t hold our doctors to 99.9% correct so how much are we gonna be withholding from people because we’re waiting for perfection but safety and perfection are two different things that I think they can be separated yeah yeah so that’s why we were I wanted to talk to the results in you you had a measurable and statistically valid impact on patient outcomes that you haven’t published yet but you’re breaking worldwide breaking news right now um and also you have these safety measures in place you know thank goodness they weren’t triggered but but they’re in place and if needed they’re in place and it’s not if they’re going to get triggered at some point they’re going to get triggered yeah it’s not if it’s when and and we will watch you to watch that very closely anything else you wanna uh talk to with robot no no that was fantastic update great yeah I’m super excited I mean you really were a leader and instead of bringing this out in the best well organized way possible but still there’s no certainty and it’s really excited to see it having some some early early success so now what we’re not over the finish line yet uh we we have to take it to the next level and go ahead and get um get those uh codes uh so we mentioned we had had a high level agreement with so we can show good numbers and so we’re ready to present the numbers back we are talking a little bit about them here and we do have actually a second insurer who is also expressed interest in this as well because everyone’s come to the same conclusion that we know that mental health has a very significant impact on Total Caustic Care we got to do something and we think this is a good solution it’s not the only solution that needs to be in toolbox but it’s something that we strongly believe in yeah excellent well now I wanna broaden now to talk about the overall um AI market Loms um do you have a minute just to sort of talk through what you’re seeing in the overall market you’re you’re sort of um one of our AI experts that uh coming in to help us understand the overall space a little bit you know when the previous previous session that you put out the title was AI Pioneer I am not an AI pioneer but I’ll take expert for today so I’ll try to give some insight as to what I understand I’ll tell you what I mean they talk about exponential growth being hard to keep up with we are it is hard to keep up I am spending a ton of my time just trying to understand this stuff and I get to do this all day um it’s it’s it’s it’s wild but very exciting but yeah let’s let’s unpack it yeah yeah so um I wanna now talk about the large language models and mean they are growing and changing and evolving quickly and we’re starting to see some healthcare focused lolms and what do you see now there how do you evaluate these models what should the listener look into uh that really would be applicable in in medicine or in healthcare and pay our markets way I I’m trying to think through some of the large language model use cases in healthcare it is well and and again I think we started a little bit about this open versus closed model so first question to unpack that a second the closed model is you define the all the parameters of responses and even though it can listen broadly it will answer narrowly and my initial mindset was why I want you to answer narrowly the in the downside of that is you start to lose engagement of the patient if it sounds like the response is always the same thing and a very contained very it’s controlled but you may not get engagement and tell a story of I try to play golf I was taking a lesson and I was slicing the ball and kept on doing it no matter what the instructor said and then by the sixth time it started straight now and I curious what would you told me to do if I kept this didn’t work he says I don’t know I only have six ways of saying the same thing hahaha but a lot of times in healthcare you have to adapt your language just like you know what you’re not this is what I’m saying is not registering with you so let me try this angle let me try that angle but the principals are the same it’s not that the principals aren’t new I’m just trying to convince you as a health coach a lot of medicine is being a health coach how to adopt and and change how you do these things close systems are going to eventually struggle with that so the everyone who has a close system recognizes the LLM is coming the question that people need to be asking themselves are is the LM going to have open data or coming pulling data from a public source or is it only going to read from a very narrow source and if it’s an LM that reached from a very narrow source we don’t really know how that’s gonna play out because these things consume enormous data sets right right right right so there’s this conundrum that’s to be determined but I a question I think people should be asking themselves as they’re betting this is so what are you reading from what what is feeding your model and if they can’t tell you what is feeding your model or you need to be comfortable with feeding their model and in turn in order to some of these better or empathetic responses they’re gonna have to read from other sources I do feel better about the hallucinations say more about that why do you feel better do do you are you seeing the hallucination rate come down significantly are you seeing guardrails but it being put in place that you know what what’s the real progress happening there cause this is the first time I’ve heard anybody say they feel better about it didn’t say I feel good yet I just feel better better better better better yeah the reason I I’m starting to feel better about the hallucinations is an understanding and I’m I’m probably outside of my depth on this one is what we call temperature in in LLM models so LGBT is generally a warm model which allows it to have a lot of creativity and you want to have creativity right cause it’s made it can make a rap it can make a I am a contameter sonnet but in healthcare we don’t want it to be too warm we wanted like if you don’t know say you don’t know now that we’re able to start getting the some of these more healthcare focus models they’re tuning them accordingly to healthcare and that’s where I’m starting to feel a little bit more comfortable if you still ask you can probably trick GBT 4 0 to hallucinate because it’s a warm model hmm but for some of these healthcare specific ones that were starting to toy around with that’s actually pretty decent um response is what excites me about them is because we’ve been predominantly using clothes models very programmatic machine learning or knowledge graphs but the speed to adoption to these or sorry the speed to programming these things is so much faster now before it would take us a year 6 months to a year to really kind of tease out a knowledgegraph model now coming back and I like I’m playing with this thing within a couple hours of it again there’s a big difference between though playing with something and feeling comfortable from a safety perspective but I do feel that there’s a lot more conversation happening out there about safety platforms whether be um you know social work that’s being done by Justin Norton and and and and the qualified AI team Hippocratic AI has put out a safety portal where you can test out what it what it’s doing uh there and and this is starting to to uh to expand so the healthcare LLM is starting to make some make some progress here yeah and on that temperature um lever I I was trying to start up I didn’t end up investing it but they were trying to bifurcate the like the social cultural um small talk aspect of the LM for lack of a better word and then um make that very warm creative epithetic but then kind of walk down the clinical pathway recommendation piece so wouldn’t be um create creative or open much more science evidence based yeah I don’t know that they have figured that out but but we’re starting to get more and more um use cases or more nuance around it which is kind of interesting like yes I but the fact that we’re even having this conversation that you’re able to potentially bifrogate a model now that some parts of the conversation are warm and some parts are cold and more just exacting I mean sure like that’s what we would want in the ideal doctor like someone who is a just remarkable technician and just focused for when they’re in the or and then someone who’s you know Trapper John MD when they’re at the bedside I’m sorry I just completely dated myself right yeah but that’s kind of what’s exciting about this is that way we can create multiple aspects of this base off of you know short and concise and sweet to the point when you’re in the or because that’s Bam Bam Bam time is time is life and Gregari is an outgoing at the bedside when you need to be talking to somebody in a in a in a very warm engaging manner I I this is exciting yeah and so the pace of change I wasn’t that confident that they were gonna execute it but but the pace of change is a really really exciting as you said to be it’s hard to keep up with but there’s a lot of change in and this nuance meaning we’re no longer satisfied with okay can write a sonnet in four languages but okay what can we do that could be really could be more applicable I saw a meme somewhere which I think does summarize some of the frustration you know we’re gonna probably from a gardener perspective start getting into our trough of disillusion amend soon uh but I want AI to do my uh laundry and dishes so um I can do art and writing not it you know the other way around right right right right yes yeah cause yeah and I think that’s what happened you may feel may have seen the gym I add that was playing during the Olympics um about you know show my daughter how to write a letter to write a letter to my for my daughter to her track star idol yes yes and they and they pulled it because wait that’s not what we want AI to be doing no no no I do think that that was that was interesting so it feels to me like next year this time we could be as we have you back on we could be having a conversation that has nothing to do with the technical ability of the LLMs you know we we let’s just assume in a year we’re there and it’s much more about uh the regulatory environment and society’s willingness to adopt um what what are you feeling or seeing there I mean part of your job as a you know chief digital transformation officers to track all the tech and and run these pilots but part of it is to be able to communicate and translate and also track the regulatory environment and how ready the rest of the world is for where the technology is where where are you seeing things I mean obviously the LMS are moving at a breakneck pace and there’s been so much capital whether we’re talking about the magnificent Seven or venture capital so much capitals been pushed into this space over the course of the last year um what are you seeing outside of that what are you seeing in the government what are you seeing in just you know your co workers willingness to sort of accept the fact that this tech is really starting to get to the place where it can be truly helpful to them I I I think that’s actually that is the question uh that I am trying to spend the most of my time on trying to help my organization understand what is the impact I remember when GPT 3 3 5 came out and I did a presentation on board I put a slide together and one of slides walls 1 slash 99 everyone’s gonna focus on the 1% do I have an update of scientists to build out my text stack and all that and I think where the opportunity is like which organizations can figure out how to get their 99% of their workforce who’s not in the deep in the text stack to understand and adopt this and the hypothesis is completely born out there’s still a lot of excitement and hype around the 1% most organizations whether your healthcare delivery organization health anywhere in the space you need to be thinking hard about the 99 is what I I thought folks I think we are going to have some really interesting challenges around identity in healthcare and what I mean by that is there are tasks that these tools are starting to be able to do but these tasks fall into historical identities within health healthcare rules that are court to the identity of that healthcare rule so let’s say we talked about robot having a conversation in an empathetic fashion has been core to caregiving for a long time we’re not taking that away from the clinician but there is but there’s another participant in the conversation now and so there is going to be I think increasing concern from well what is this affect my identity I think a lesson Learned for us here and it may be too late it may be a situation where it’s equivalent of Kleenex is Kleenex and not tissue paper and Xerox etcetera virtual nurse I think may have been one of the worst terms we could have come up with because there is nothing virtual about our nurses who are interacting with these patients they are real live human nurses they just happen to be using promoting technology so they can come into the rooms but guess what there are some remote there are actually some virtual nursing tools now that are truly L O m E I and and by the way we talked um I I know if you saw um the video clips uh when were in video when they had um they talked about democratic AI they had the virtual nurse with the yeah with the Avatar and they got a lot of flack for that interesting line up rumor is that wasn’t Hippocratic AI’s idea that was videos made them made the video for that so because India thought it was cool right they don’t know healthcare yeah but then the healthcare size like dude that’s creepy right right but what’s really great about some of those tools the voice interaction is great it’s really it really seems pretty solid uh so now though we’re gonna have this buck butt up of wait am I a virtual nurse am I or is that a virtual nurse or am I actually no a nurse who’s using tell a technology to remote in such that I can you know be more accessible cause I I can be in multiple hospitals and covering multiple hospitals same thing with our clinicians and I think you know we talked previous before we were betting big on this we’re wiring our entire health system every met surge room by our goal is by next quarter sorry first quarter 2025 we’ll have um you know we’re on the carry I platform and big news on that one too um but you know we’re betting big on the smart hospital because we recognize what a limited resource our clinicians our physicians our nurses etc are we’re gonna give them every tool possible to spread their presence into patients rooms ideally we want to be in the patient room too but at least we’re gonna enable them to I don’t have to have a cardithractic surgeon driving between hospitals and you’re in the or and does a video visit with you when you’re you have an abnormal catherization and then we’ll meet you at the hospital you’ve already met your surgeon in advance that’s a big win for us so but now we’re starting to get into identity questions and I don’t have the answer to this this is what I’m working on and thinking about um it’s gonna get interesting yeah and I agree it’s the question it’s also a marketing branding uh ego question not really a technical question right through um I in my own mind I’m a great driver I could have been an indie driver but I decided I wanted to hang out with Marcus and do venture but I still have Andy lock brakes in my car and there’s no way that I’m gonna be able to pump the brakes at that pace per second in on a wet road and even even like the most egocentric person like me that thinks I’m a good driver you still want interlock brakes right so that we got to figure out how we can say to the physician the nurse the administrator these are tools that are gonna empower you to do the things you really do well and then there are things like in like breaks that just a computer can pump the brakes whatever it is 4,000 times a second and even though I am in my own mind a good driver I can’t do that right so it’s not it’s not like diminishing to me to use these tools we have to figure out how do we position it like that so that our great clinicians wanna you wanna wanna use it and I I think then the question comes back to what been using the race car race car driver analogy is the feeling well my identity as a race car driver is being able to drive the most raw form of automotive technology without assistance and I think you can potentially look at Formula 1 now well we still think those are the greatest drivers in the world right they not even drivers they’re pilots at this what they call them yeah and they’re able to manage that technology in the cockpit at any given time and yes they still need to know the lines but they’re those drivers of today versus the original drivers the one manual Fangios etc different drivers and it’s okay yeah and but yeah and in some ways better I mean they’re definitely different but they’re they are very proud of the skills that they have today they’re very tech enabled but it doesn’t diminish them in any way I mean you could just look at the steering wheel is two radically different steering wheels right right the number of buttons on that they’re having live conversations with their pit while they’re driving 200 miles an hour you know around a turn I mean that is a different skill set but they still identify themselves as race car drivers and I think that’s the challenge that we were gonna have to have in front of our clinicians to say you’re still a clinician even though you’re not the purist with the goggles uh in the race car anymore you’re gonna be monitoring all of these other systems while you’re still steering the patients help to an outcome if you want to say it well I want to drive old school way and there’s a place for you but that may not be in the modern healthcare system and I think we can help get our folks there yeah TK I actually just texted Bruce Brandis I told him we’re talking to you it’s time to get him on the show but as a preamble to that help thick and I understand we were talking before we started we hit record help us understand the significance of the striker acquisition of Care AI so full disclosure biased here where we we’ve been working with Carrie I we as a man till we’ve committed to wiring all of our Med Surge and ICU beds with carry eye sensors we’re about 300 or so in and but won’t be done by first quarter of next year the argument behind this is to me a big swing at the concept of the smart room and this is not in virtual nursing play this is a are you are you believer in the smart room the smart hospital room of which virtual nursing which again I think that’s a debatable term is a component of it but if you look at virtual nursing virtual nursing for the most part actually doesn’t remove nursing uh it actually adds and we’ve made a you know as an organization we’re not opposed to that because there it’s bringing the right type of care to the to the bedside human to human of course with a telemediation in between where you get the wins to pay for this is all the other ancillary technology or redundant technology that you don’t need in the room at the same time so today if someone needs to be monitored for telesitting you roll in a separate carton camera for that now you don’t have to do that now you know the I think the synergies between striker and carry I now is well striker makes beds carry I has a sensor now Carrie I can look at and say you know what that patient is starting to move a lot normally or they’re starting to move through those thrashing around a little bit let’s turn on the camera and see what’s going on with them their ventilator it’s time to wean them off the ventilator let’s look and make sure that the bit is at the right angle so there’s a bunch of synergies that start to come in here to make the heart the room smarter and you know I think it’s a I think it’s a pretty smart move for a striker to swing big on this and now you get carry a bunch of funding on these and in use cases very um very very interesting time ahead yeah and what what I want you just to spend a minute on is uh kind of play out the the strategic road map I mean the combination is really interesting in that um that AI intelligence layer with strikers physical hard in a hard products is a pretty powerful combination out there and I haven’t seen that combination put together anywhere else mean there are other makers of beds and devices and infrastructure there are tables and everything uh but I don’t know that any of them have an AI layer um is that is that kind of your view or or who else do you think is in this space that’s gonna um it’s gonna wanna buy someone Marcus in my portfolio this is already you know haha so from an AI let’s say ambient sensing wait the way I’ve been thinking about it is that from from a sensing or an AI layer you can either watch optically and visually or you can watch ambiently with using lidar technologies or or or infrared etc the legacy players in the space watch with camera bandwidth the problem is you can’t watch 1,000 rooms in in a health system or a medium sized health room system you can’t watch with 1,000 cameras at high definition you’ll you’ll destroy your bandwidth right it’s there’s not an option to do that so you got you gotta be able to to watch ambiently with shapes and computer Edition which is and and in the other pieces you got to be able to do it at the edge where the entrance computing is happening at the edge not all at the central point of it right but now you’re really starting to get into the concept of is micro internet of things and and so the internet of things is the well what are the what’s the IoT inside the room and you need to start getting into a critical mass of devices that are actually able to feed off of each other because if let’s say the carry I platform detects that normal movement now the striker bed could say yep I’m seeing it as well or no that’s not abnormal movement because that is a family member who just put their bag on the person’s lap and that’s not abnormal because the bed is able to detect the difference so as we’re able to understand more and more believe it or not one of the hardest things for us to figure out in healthcare is how much weight a person has lost or gained during a hospitalization well if I have multiple sensors now telling me I think the weight is this then I can adapt my treatment protocols cause you get admitted for heart failure my goal for heart failure is for you to lose X percentage of your body mass which is excess weight in X number of days and if I’m not doing that fast enough I’m not doing my job problem is weighing patients is a root pain right and very inaccurate but if we can start figuring other ways volumetrically what is your weight what is body mass stretching look like what is impedance look like now we have a very different way of managing people so I think the smart room is only gonna get smart and do those uh sensors have to flow through the EHR or can they be separately I don’t think you’re relying on the EHR necessarily is that true so they can get away from epic Cerner uh bottlenecks at some level you are not reliant on all of it running through your EHR absolutely the EHR can feed it and you can feed the EHR but you can run these things completely stand alone and and I think where they’re recognizing there’s an opportunity to compete in this not even compete in the space but I think augment in this space is there’s a lot of activity that can be captured in real time both Indian listening but I actually even think now through the concept of ambient vision you know where there’s a technology couple technology we’re using that will watch a procedure to I think we mention there’s augmented reality during a colonoscopy where the the scent that it’ll put a box around if it thinks it’s a polyp which is right and I would argue if you’re gonna have a colonoscopy you better ask and make sure that your doctor is using one of these ambient technologies because a clinician can miss it but the other thing I would say is why can’t this thing generate your off report afterwards it watched the procedure and now it’s able to do that well you know what that gets immediate adoption so why not have that conversation have the entire conversation with the patient in the room and then by the time you walk out they already forgot half or didn’t ask the question they wanted to now they can listen to a transcript of what just happened you know if they’re if they’re interested in that if they say I don’t want that that’s a different story yeah right or they if they approve a family member could could be let in to see that as well cause they have a lot of questions and it’s not that the mom or the grandmother doesn’t doesn’t allow the filler but they can’t get the doctor back and they don’t recall that I know we’re probably done with time but the use case they have for us that completely sold us on this was family member got it admitted um elderly gentleman wife was at the bedside I said what’s the medication list and we don’t have the medication list oh but our son in this state has medication list added them to the call on the flight okay what’s the medication was oh you know what my sister picked that said she picked that up let me get her oh no no problem text she’s on the line and now we had four people having a conversation around you know the patient’s information we got real time information got them the information you need it try doing that in real life how often do things get missed because you can’t be there so I think that’s where this is the reality of the world we’re gonna have to be in yeah so let’s uh we are coming from time let’s just spend three minutes on where you think the future of the workforce is headed I mean this is obviously we’ve already talked about it gonna really impact how we kind of organize our human labor to treat patients with the most most effectiveness and best outcomes for the lowest cost I know you’re really involved in thinking that through how should the audience start to think about that or give us just some pieces and parts that we can start pretty and we’ll probably come back to this at another time but how do we start beginning to think about the future of work and healthcare oh and and again I reference um my uh my friend Khabi softly beyond this one and he and I think you know some of the things I’m gonna share he’s helped me think through this as well is we know already that AI to begin with doesn’t replace jobs it replaces tasks but then if you start to replace enough tasks the work that’s left behind is fundamentally altered not replaced altered so now the question is what is the type of person who can be way more flexible and do one thing today and a different thing tomorrow and that is very different from the skills based world that we’re in we think about today the skill set as gonna be needed for the workforce of the future is going to be the critical thinker yes we have lots of critical thinkers in healthcare but it almost at every level we’re gonna need critical thinkers I think that’s what we’re trying to figure out is how do we understand the skill set it’s very easy to know whether someone has X y Z skill set we can their verification programs to do that but do folks have critical thinking skills because the machines will make mistakes and so critical thinking will say oh it doesn’t right critical thinking or I can do this differently critical thinking so you have to be thinking through how am I going I and I think you can teach critical thinking too it’s not an inherently you know talent but what are people doing to not just teach skills but teach critical thinking skills and that needs to be built into your workforce development plan at a very early stage and I don’t know how to do that yet but figuring that out yeah I mean I I kind of think about that as as as you know finishing skills or you know quality assurance skills and most people are geared towards you know the first 80% of the work not the last 20% of the work you know and and are actually pretty sloppy when it comes to the last 20% of the work right and and so that’s um that that’s a fairly big shift and I think there’s an identity shift within that right because I think most people are very anchored to their strengths and their values being in that first 80% of the work and switching them to being finishers and quality assurance people uh that’s that’s gonna be interesting yeah and our existing workforce training systems from high school to college to professional schools are really focused on those skills based largely and then they sprinkle in a little bit of critical thinking or strategy work as a nice to have right we need to maybe flip that um so I mean that this has been a lot I think that that’s a good place to leave me just even that that concept of AI doesn’t replace jobs but it replaces tasks and as you start to replace multiple tasks the jobs that people need to fill is gonna inherently change that is really good I need to think through the repercussions of that but yeah that’s there’s a lot there’s a lot there it’s gonna be interesting few years yeah I mean so just hit I mean the last hour we have this bolus of the boomers which is just now coming into their 300% utilization peak we’ve hit the midpoint but that means there’s half more coming the tool year we bought testing is going really well you’re gonna publish the reports internally but the summary is going really well and you’ll keep expanding it I hope and the Loms that more open systems are starting to get much more powerful and we’re getting to have more more levers to be able to sort of pull in the the temperature in some areas and so that maybe I don’t know if it’s fine tuning or or designing it for healthcare uses uh huge acquisition with striker and Krai that you’re an early adopter in and you’re rethinking the entire way our workforce changes that that’s a lot to digest in an hour and really thank you for talking us through a lot of change in the next 8 10 years um and I’m pretty excited about it I mean there’s a lot of positive some things to think through but for the most part it’s gonna be better for patient care better for outcomes agree I concur as well I concur as well so it’s just it’s it’s making sure we’re having meaningful conversations along the way and and being very open and transparent with each other as to what it’s gonna look what we think it’s gonna look like and just not giving false hope or promises but also not sugarcoating it either is oh no everything’s gonna be the same because it’s not so TKC in six months you tell me when I don’t have all right news in six months but we’ll go home the last six months you produced a heck of a lot so I think I’m really excited to continue our conversations if something breaks let us know but but the law I really appreciate you on the inside of a big health system feeling willing to you know think about things on on the edge and and really try thing in a safe way but but but try to figure out what’s next for our for our whole industry appreciate that and we’ll uh maybe see at the 3 p yeah there we go yes alright thank you buddy take care Jens