17 Years Behind: Dr. Ido Zamberg on why hospitals can't get their own best practices to the bedside

0:00Hey, Tim, good morning. Hey, good morning. So I was in clinic yesterday. 12-hour day. I'm going to never take vacation again. It was pretty brutal. You're too old for that, man. 12-hour day. Yeah, the real work. Sounds like you've been hustling, too. Yep, and been celebrating the Carolina Hurricanes. We're a hockey state now, Tim. Oh, congratulations. Do you know the rules? Yeah, you know, you put it in the net. You get a goal. You just push reports. You yell when that happens, right? Yeah. I still don't understand. I used to go to Richmond Renegades games when I was younger, and that was fun just because of the spectacle of it, but still never learned the rules.

0:40Watching it live, it's an amazing sport. Oh, well, I'll have to make a trip down there sometime. Should we get this everywhere? Let's dive in. Let's skate in.

0:52Welcome to Health Tech Remedy, the show where three physician leaders in health technology break down the stories behind today's most innovative health companies. and we speak with leaders building the future of care. I'm Trevor Royce, radiation oncologist and researcher focused on real-world evidence, informatics, and AI diagnostics. Joined by my co-host, Tim Showalter. Good morning, everyone. I'm Tim Showalter, radiation oncologist and former med device entrepreneur, now focused on scaling AI technologies that improve patient outcomes. Today, we are talking about C8 Health, a company focused on a problem that essentially every clinician will recognize immediately, even if we do not always name it clearly, how hospitals manage clinical knowledge.

1:32C8 is not about trying to diagnose disease earlier or replace doctors or automate clinical judgment. It's really just trying to make sure the right clinician can access the right institutional knowledge at the right time. On surface, maybe it sounds like a little simple, but it gets at one of the great paradoxes in healthcare, that medicine is incredibly knowledge-intensive with guidelines, protocols, pathways, policies, order sets, quality metrics, onboarding materials, your own specialty society or disease-specific recommendations, local institutional rules. But when you're actually taking care of a patient, that knowledge can be scattered everywhere.

2:11PDFs, intranet pages, intranet pages, SharePoint folders, email threads, binders, paper copies, EHR links, and so forth. But the memory of whoever happens to be on service that day and on and on and on, how do we solve for that problem, get everything in one place? Yeah, I would also add on bulletin boards, I still have the clinical space that we work in. But the consequences of all this distributed siloed data is not trivial. So the problem is that when it's hard to find knowledge at point of care when you need it most, care actually ends up becoming variable. And variability is generally the enemy of quality.

2:49So this means that quality initiatives that the hospital is trying to do or departmental leadership is trying to do often will not translate into actual changes in practice. And then new staff have to relearn local workflows. Trainees come in and out of services. You know, old dogs like us have to learn the new tricks as things change. And then you've got these like travel nurses and things like that who may be clinically excellent, but blind to local operational standards. So it's not just an inconvenience problem. It's really a quality problem if you think about it. It actually can be a safety problem too.

3:25It can be frustrating for the workers within a hospital. And it could even be a reimbursement problem when practices are not aligned to institutional expectations. yeah i think the bulletin board comment you made really resonates with me because it's so true some pieces of information you just know are going to be right there where you need it and it's on that bulletin board in clinic and who knows how long it's been there updated and like to your point just to connect the dots a good example would be like clinical trial often you'll have some clinical trial that's been opened and you need those criteria or the details about that trial okay let's print it out and put it on a bulletin board but then it's not updated you don't know if those change or if that trial closes and that person that may be running that trial may go on, but their, their, you know, number is still listed on this sheet of paper that's been printed out.

4:15It's, it, it's almost hard to articulate how true that is. Like this knowledge is truly scattered everywhere in the healthcare system. Completely. Or, or like if that, the, the protocol was amended and then you've got to like tack the new printout onto your bulletin board. That's, That's challenging. So, I mean, maybe this is a good transition to talk about what C8 actually does, because I think this is where obviously they're working towards a better solution. So C8 Health describes itself as an AI powered best practices implementation platform. So I think that wording is key. They're not just simply saying we're like a storage for knowledge base, but they're actually saying we take action to help hospitals implement best practices like that.

4:59that sort of follow through is a really important distinction because the knowledge base would just store information, but they're trying to actually push that into the day-to-day point of care work. So they're pulling together clinical protocols, guidelines, operational workflows. They're taking the thing off the bulletin board and putting it there into the platform. And so clinicians can obviously access that wherever they are. So mobile, desktop, EHR linked workflows, And then they've got an AI assistant, Panda AI, that provides access to that information in plain language Q &A, essentially to reference the institutions previously vetted and approved materials.

5:38So it's basically not like what can a general chat bot do, but the question is, you know, how can you access your hospital's approved information soon? Yeah, and that Panda AI, as you mentioned, is the tool that makes all this happen. But the key in all this is that it is local data elements or sort of locally oriented. And what I mean by that is, yes, there are these national guidelines, there are national references for best practices, but at the local, that local level is still critical. And a good example maybe is antibiotics. You know, antibiotic prescribing patterns are often driven by sort of local resistance patterns on best practices for that area.

6:25Or, you know, national care guidelines on the cancer side allow for a lot of different options. And it may be that some of those options are preferred locally based on, you know, your own institutional infrastructure and resources. So that local practice piece is actually critical. What's the protocol in your given hospital? Which pathway does this department use? And so forth. This is where the national evidence really meets the operational reality, boots on the ground, your local system. Yeah, and I think that's key. And obviously, like we've seen over the years, a lot of consolidation within healthcare.

6:56So there are these, you know, health systems, like even, you know, at my own hospital, regional hospitals have been acquired or groups have been added to the practice. And ultimately, if you stop and think about it, hospitals and broader health systems are not knowledge organizations. They're not really designed to house new information and enforce availability of that information. They're good at care delivery. The last mile, they're good at sort of following regulations on the hospital side and billing and training residents, obviously. But knowledge management is not a core capacity you think of for big hospitals.

7:37And obviously, we've talked about this before on this show that the EHR does not solve this dilemma. It's good at storing patient level data and supporting transactions, essentially. It houses that data. But in terms of bringing forward institutional knowledge, all of these different data sources and sort of knowledge documents that we've talked about, it doesn't need, that's not like part of the system. And so I think that's a real limitation. And when you add in the realities of like different people rotating through lots of different specialists involved, new locums, folks, old people like us who need to be, or like me, I should say, who need to be caught up on the latest thing that has changed.

8:23It's really, you know, the hospitals generally lack support for that. And I think this is where C8's framing is really strong. So they're They're basically saying that not that the documents themselves are disorganized or out of date. It's that the challenge is a handoff into the actual point of care. I think that's where this performance management piece becomes so important. Yeah, absolutely. An interesting core concept here is that knowledge fragmentation is true even within the EHR and all sorts of new tools emerging that help bring the EHR knowledge in one place, like using large language models to search the entire EHR.

9:01But this is really up-leveling this. This is beyond EHR across the organization. So let's take a brief moment to discuss the origin story of C8. And that starts with Dr. Zamburg, who was a programmer turned physician with a background in database engineering, and then ultimately became an anesthesiologist. So a physician-led perspective. I think we'll have to hear from the C8 team firsthand, but this gives him the lens that sophisticated technology companies manage structured and unstructured information really well. Clinical medicine's not doing that. How can we bridge that gap? Yeah. And I think, obviously, like, as you've been in ORs a fair amount, I think anesthesiology is a really logical starting point.

9:47It's very, like, high acuity, need to have information right their point of care. There are key procedural workflows, you know, handoff safety protocols, and there's constant coordination across teams. And so it just lends itself really well to having policy-based knowledge and guidelines translate into real patient improvement. So I think really, really high impact in obvious area. And I think it's from a commercial standpoint, I can imagine for C8 Health that starting in the anesthesiology place would make a lot of sense from a go-to-market standpoint because it's pretty operationally complex and it's in like an OR and inpatient setting where that can be often really important for hospital administrators.

10:31But also, you know, it's interesting that C8 is now expanding beyond that area pretty clearly. And so there's a recent VentureBeat article that frames the company as moving to whole hospital knowledge management. And I think that makes a lot of sense from the company perspective. Obviously, this applies in internal medicine and surgery, emergency medicine. You know, it's obviously very applicable on the hospital side, on the inpatient side. And so I think this is where there's a real natural fit for expansion. Absolutely. Let's talk a little bit about the AI component here and why AI is useful in this case.

11:07You know, it is a different use case from a lot of what we think about AI in healthcare, where you're using it for different ways of diagnosing patients or predicting risk or accessing, you know, the EHRs we mentioned or interpreting imaging like screening lung cancers, AI detection of these nodules and so forth. And that, you know, that's pretty buzzy and exciting components of AI in medicine, applications thereof. But C8's a little different. Can you tell us, Tim, a little bit your thoughts there on the AI component of C8? Yeah, I mean, I think that this is a pretty clean use case of generative AI in healthcare.

11:42So you're not inventing things. here everything would presumably be traceable and it's a good example of you know creating a curated sort of walled garden of of content and using ai to pull from that so it makes the accessibility really easy of course very often the clinician would still need to prompt for for the information but i think you know it it makes finding that information quickly and going straight to it really easy. So rather than looking for page 37 of the particular PDF, I think it really makes it helpful and can link you to the direct source, which is really helpful. And I think this is sort of an interesting middle ground of safe AI that matters.

12:25So I think I'm pretty bullish on it. Maybe for our listeners, before we move towards the interview, the other thing for us to think about is where this fits into the healthcare AI landscapes. Trevor, do you want to just cover quickly the competitive landscape in this area? And when we think of competitive landscape, you could call out a couple of different entities doing some interesting things, but it's almost more helpful to see what this isn't. It's certainly not up-to-date or Dynamed or sort of generalized medical references. It's also not really the EHR, obviously. This isn't the documentation tool for your clinical encounter or your clinical transaction system.

13:08It's not really a learning management system where you're, you know, doing training modules and so forth for educating your workforce. And it's certainly not a traditional clinical decision support where you get alerts or rules-based ordering pathways. It's basically a clinical knowledge infrastructure layer across the organization. So it's kind of like an enterprise knowledge management, that might be the best analogy, but specifically for healthcare and its operational realities. Yeah, I think it's spot on. So it's not really exactly like the companies you might read about. And I think there's some strategic questions to think about, like how this is operationalized.

13:47Does it remain standalone or is this something that would be really well suited to pulling in within an EHR ecosystem as a sort of smart app within an EHR obviously no one in healthcare wants to deal with another login or another app or something that's outside of the EHR but obviously the EHR has not solved this problem well and so we do know that there's there's something there and this is obviously a step forward beyond the the printout on the on the bulletin board in the workroom so I think it sounds it seems like it's integrated enough to avoid being a single silo but it still remains pretty independent from the EHRs and it's interesting to see how that evolves over time.

14:29I think that you want to make sure that nothing requires an unnatural step for clinicians. So obviously that's important. Well, I think we've set a table nicely here. Should we transition to our interview with Dr. Zandberg? Let's go for it.

14:48Our guest today is Dr. Ito Zandberg, co-founder and chief medical officer of C8 Health. Dr. Dr. Zandberg has a very interesting background that spans both computer programming with experience at large software companies, as well as medicine and now health technology. So we're really excited to have you with us, Dr. Zandberg. Welcome to Health Tech Remedy.

15:08Ido Zamberg: Thank you so much for inviting me. Great to be here. So I'm happy to kick things off. I mean, I think it would be great to hear a little bit about your origin story. You know, I think you have a very unique background and it makes sense when you read about it, about how you ended up tackling this. But can you just sort of share kind of your career arc and how you came to tackle this problem that we'll talk about today? So, as you mentioned, I'm a physician. I trained in anesthesia and generally internal medicine. Clinical passion is OB anesthesia. I also spent many years in medical education. I have a master's in medical education from Johns Hopkins and was the vice chair of clinical competencies education in my faculty.

15:47Ido Zamberg: And finally, I'm a computer scientist. So I actually worked in the industry both before and during med school as a senior developer in large multinational companies like HP Mercury, Autodesk, where my expertise was optimizing access to information in large data sets and sharing of information within the enterprise and between the enterprise. So that's what's got most of my career. And then the last few years as a computer scientist, I was in mobile development, a really front end. So both from a clinician perspective, I really had a variety of experience, both from the super chronic, so rehabilitation to super acute in the OR, emergency care, pre-hospital care, ICU.

16:30Ido Zamberg: Then also he's a computer scientist from a lot of years of back-end databases, etc., through later years front-end design and mobile development. And I would love to hear a little bit about your transition to C8. I mean, was there a moment where you had an acute recognition of this problem, where this information challenge in healthcare that you were trying to solve at C8? Tell us a little bit about that and how it translated from your clinical background at that point. Yeah, so I remember, I mean, the exact moment I was, you know, as a medical student, you have, you do rounds in different places, and I was in my geriatric rounds.

17:06Ido Zamberg: And, you know, when a geriatric, so when an elderly patient comes to a hospital, right, they might have, they might come because they have a pneumonia, but then they have probably many other problems like they would have diabetes or they would have a pressure ulcer or they they would have i don't know other five or six other problems so when we admit them to the hospital you will need to do what we call a head-to-toe examination right you would look at the main problem but also try to identify whether they have secondary problems then identify you know older medications and new medication if they have you know kind of contraindication, etc.

17:41Ido Zamberg: So I noticed that, you know, every one of my colleagues would try to draft checklist and try to understand where the protocols and what type of examination they need to do. And if they find a problem, what's the stream? What's the next step? What do they need to test? And I was just looking around me and seeing that everyone is doing the same work, but very differently with a lot of variation. And I just didn't understand why, because I saw, because I knew that the hospitals had pro goals with exactly the right instructions of what we need to do in a specific site where we work. That was just weird for me.

18:16Ido Zamberg: And then as my job for many years in the industry was making sure that people can reach the information they need, I realized that people are not using the pro goals because they didn't know how to find them because we had multiple systems and shared drives and intranet websites and even protocols that are stuck on the wall that you don't know from what year they are, if they're still the latest version. So it just was upsetting to me because my entire career as a computer scientist, my whole expertise was optimizing that access to relevant information. And when you go to healthcare, it's such a knowledge intensive domain.

18:56Ido Zamberg: And I was just very surprised that we don't have a proper system to handle all the knowledge that you need as a clinician. So I remember sitting down on a bench in the geriatric hospital and thinking, why don't we have a better solution? So as a first step, I tried to kind of look at what exists in the industry, right? We have, I mean, in the tech industry, you have softwares like Confluence and Asana. You have a lot of knowledge management systems that are specific kind of for the tech industry. and I tried to deploy it in my department, but the user experience and the functionality doesn't really add up.

19:33Ido Zamberg: So I just sat down for a few weeks and just really created something for myself egoistically because I wanted to have access to all this information easily and to do my job in a better way. So it kind of started like this and then I deployed it in my department and then very quickly it spread across the entire health system, across five hospitals in the entire, all of the departments, and kind of started like this. And wouldn't you say you created something? Can you tell us a little bit about what that, you know, first iteration looked like? And then I'll pass it over to Paul, who I think will talk a little bit more about this big problem in knowledge management.

20:09Ido Zamberg: Yeah, I think from a high level, I mean, maybe I'll tell what CA does, but we are a system for implementation of best practices. So the idea is that the hospital, as I briefly mentioned, I mean, Every hospital has a set of policies, procedures, protocols, guidelines, a lot of materials that as clinicians we need to know about or to know of when we perform our job. The thing is that they're spread across multiple places and it's not easy for us to find them and to know where they are. So really what we do is we provide a system for the health system for the hospital to help them manage and disseminate this information in an efficient way to the clinicians.

20:51Ido Zamberg: And then from the clinician perspective, one single unified platform where I can find any type and form of information that I need in order to perform my work. Yeah. So, you know, I mean, you've sort of talked about we have information available somewhere. Then we have seeing patients at the bedside. And it sounds like what you're describing is a situation where normally these two pieces of information kind of live in different silos. It would seem like it's the kind of thing where, you know, naturally, 30 years ago, somebody should have been saying, how can we move the data and the information from one silo to the other?

21:35How can we bring this guideline information, these other critical pieces of information and patient care to the bedside? Why do you think that's been so difficult? And then I guess, what do you think is different about C8 that you're tackling a problem that I think is not new, but that hasn't really been effectively tackled before?

21:55Ido Zamberg: So when I was a computer scientist and then I moved to medicine, people were looking at me or why are you doing this? I mean, it's a completely different world. It's not the same. And then as we move forward, you know, with the years, the tech industry really penetrated significantly. So healthcare. So I think it's just a question of moment in time and the right technology. It's been years that people have tried to solve this problem with, I don't know, palm pilots at some point and an intranet website. I think what we do very well is that we're all about evolution rather than revolution. So we're not changing the clinician workflow.

22:39Ido Zamberg: We're just improving it and just improving whatever they do today, just helping and do it much easier and better. So I think that, and I think that's, I think the way it's been done in the past is that tech companies, and I look, for example, I have a, there's a medical device that we use in the OR that is 100 % was not built by clinicians because it's a medical device that is set for a massive hemorrhage. And then basically the three different models that it takes you around 10 minutes to build. So it means that when someone is hemorrhaging, You don't have 10 minutes, right? You need something that you just plug and play.

23:16Ido Zamberg: So I think a lot of the technologies that was done in healthcare was done by non-clinicians. So when you don't understand fully the problem, it's really hard to build a relevant solution. So I think that, and I don't want to be pretentious here, but I think that just there's not a lot of clinicians that also have worked in the industry and have seen different workflows and have seen how things can be done and really understand the problem that are trying to solve. So I think that there I was lucky to be able to really understand the problem and also have the motivation to actually and the capabilities to actually solve it.

23:47Ido Zamberg: So I think it's just a matter of luck to have lived both worlds in that sense. So, you know, I think you just touched on a couple of different things there. I mean, you know, well, the reference you made about hemorrhage and, you know, something time critical, that sounds like a big patient safety issue and designing products to be ready for the real clinical problems. But you also talked about, you know, evolving clinician workflows. And one of the things, you know, that we're seeing in healthcare is challenges with burnout, which, you know, is, of course, very much a workflow tethered thing.

24:17If people are struggling in their workflows, that tends to be what leads to burnout. You know, so I think, I mean, obviously knowledge is good, but, you know, could you give us some sense as to what you think are the problems we are seeing today from hospitals not managing clinical knowledge effectively? And, you know, kind of if you were dreaming, what do you think you're really going to solve in the healthcare system or, you know, evolve in the healthcare system by improving knowledge management?

24:49Ido Zamberg: Yeah, so ultimately, I mean, I think one of the biggest problems in healthcare is unwarranted variability in care, right? Ultimately, you know, if historically people look at medicine, there was a battle between whether medicine is a science or an art. I think we all agree medicine is a science, right? There should be one answer for a specific problem. So that's why best practices and guidelines are created. The thing is that in order for them to be followed and adhered to, people need to know about it. And that's the part where healthcare still fails at actually implementing these best practices, making sure that, for example, when there is a new guideline or the guidance change, 100 % of the clinicians are aware of it immediately and that tomorrow the next patient that they treat are treated with the newest best practice.

25:43Ido Zamberg: And today, if we look at the evidence, it takes around, they say it takes around an average of 17 years from new evidence to get into practice, which is a big lag, right? So where we sat and where we already have data that we're able to do it, our claim to fame is that we are able to increase this adherence to best practices. So when we deploy our platform, our claim to fame is that people start using the best practices of the department or the hospital. And when we look at it on a more granular level, there are also quality standards that are set that are reported to CMS or to payers. For example, if we look at patient outcomes like surgical side infections, right?

26:26Ido Zamberg: So in order to avoid surgical side infections, there are a few things that need to happen. Like, for example, you need for diabetic patients, you need to measure glucose at a specific time before surgery and specific time after surgery started. and if it's at a certain level you need to treat it and there's a long definition of what you need to do and then when we look at so it's something that is well settled in the world of very operative services and the hospitals know about it they want to improve it they know that if they improve the compliance to this metric well the surgical side infection will be reduced however when we look at the data the compliance is very low so where we step in is that we help them through our software to increase individual compliance to specific standards in addition to obviously to using the protocols.

27:13Ido Zamberg: And then when the process improves, so when the compliance to a specific process improves, well, the outcome should be better. If it's not better, it's because the process is not the right one. And then we need to change the process. And another, I mean, great example in that sense, I mean, you have a framework of ERAS, which is early recovery after surgery. It's a full set of protocols that have proved to improve patient outcomes in a set of surgeries from colorectal to C-sections to CABG, etc., which is extremely important. And it is a lot of evidence how much they improve patients' patient outcomes, reduce lengths of stay, complication rates, etc.

27:51Ido Zamberg: However, when we look at the national average of compliance, we're at around 50%, which is extremely low. So between 37 % to 55%. So our goal is really to, you know, we have these best practices. They have been proven to improve patient outcomes. Now we need to help. Now we need to make sure that people actually adhere to them and that hospitals can properly implement them with as less cost as possible and as less effort as possible. And that's where we come in the picture. You know, I think that's great to think through. It's obviously like, as you said, I think that the unnecessary variation in care is such a challenge for healthcare quality in particular.

28:31And I think what, you know, obviously on the last mile, there's so many pinch points with respect to clinician bandwidth and workflows and people just are often short on time. So with that in mind, I'm really interested to hear about how you embed within clinical workflows, like for example, how you interface with the electronic health records in particular to make sure that like when a physician's like caring, providing care documentation that it's sort of natural to follow through and make the providing good care easier? Can you give us more context on how that works? Yeah.

29:08Ido Zamberg: And I think it's a great point because ultimately what we have in mind is always that ultimately, you know, I look at myself as an anesthesiologist, right? When I have a patient, I have a lot of physical medical devices, a lot of analogic stuff. I need to prepare drugs. Technology often is something that is disruptive to my flow. It's not necessarily something that I want to do. It's something that I need to do. I mean, I go to the AMR not because it's my dream to go to an AMR. It's because I have to do it because it's part of my compliance, right? So the whole point of how we build the system is to make sure that we actually push the information to them based on who they are.

29:46Ido Zamberg: So for example, there's a flow where I can see my schedule for tomorrow, I can see my cases, and within my cases I can see who's working with me, how to contact them, what's the relevant best practice in the specific site that I work in for this specific surgery, how to set up DOR, what's the preferences for the surgeon that I work with in order to know how to set up DOR for them and what medical devices we need, how to operate these medical devices, but also I can see my quality metrics so I can see how good am I in the specific quality metric for this specific surgery? And then if I'm not good at something, well, it gives me an instruction of what do I need to do in order to be better.

30:23Ido Zamberg: So instead of a world where, you know, when we look at difficult knowledge management system that is really built on browsing and searching, the world we're creating is saying, well, I know who you are. I know what you're doing. This is the three things that you need to do tomorrow in order to be compliant with the standards that are set. That's fascinating. I'm just curious, obviously everyone asks and talks about AI. I would love to hear how C8 is using AI as part of the overall technology stack for all this. Yeah, well, I mean, I think AI, specifically generative AI, is revolutionary in the world of knowledge management, right?

31:01Ido Zamberg: Because it allows to structure things that before were very hard to destructure. So we have a variety of usages from the very core of our platform because ultimately, if I give an example of one interface we have, well, we have a sort of AI assistant that is like ChatGPT, but for your internal knowledge, right? Because ultimately, if I'm a clinician and my dream when I wake up in the morning is not necessarily finding a 30 pages PDF of how to set up an OR. Typically, I would have a question like, when should I stop a specific medication before surgery? And the answer I would like to get is the number three or four.

31:43Ido Zamberg: I don't want the history and like a seven pages output. So we have an AI assistant that you can ask questions and get an answer from your institutional local best practices. And then it defaults and it falls back to national guidelines because we are an official partner, for example, with multiple national societies. So this is one example, and the advantages here is that there's practically no hallucinations because it's only a rag environment on your content, and the answers themselves are very short, concrete, without the additional fluff that typically we see with AI. And also there's a full transparency.

32:20Ido Zamberg: It tells you exactly from which paragraph from a specific document it comes. So it's very clear. But a very important piece here is that, well, our system handles... a lot of different types of data from quality data to international guidelines to local protocols. And ultimately, as a clinician, if I'm asking a question, I want to get an answer from my institution. So also in the backend, in order to be able to do it, the entire tagging of the content itself is done automatically through AI so that you're not going to get an answer from a journal article, but you're going to get it from your local protocol so we can prioritize the relevant answers.

33:01Ido Zamberg: And then also from a recommendation engine perspective, well, also this is becoming more and more automatic than saying, well, we know who you are, right? We know what you're doing. Well, here is the most relevant thing that you need to know. And this is your AI as well. Helps us to create a lot of automation. And then one last example also from a quality perspective. So today, when you have quality data and compliance data, and you're trying to look for a problem, you're going to do a root cause analysis, right? And this is today, it's a lot of dashboards and manual work of sifting through documents, sifting through data, sorry.

33:37Ido Zamberg: And here as well, we've created models that can simply look at the data and give you as a quality stakeholder in a department or in a hospital, here's where the problems are, the compliance, here are the contributing factors for non-compliance by specific metric, and all is done automatically through our AI. Okay. Hospitals spend a lot of money on quality improvement. Do you think a lot of that is a knowledge distribution problem rather than a knowledge generation problem? And do you think this is the kind of thing where a tool like this could help, you know, not just improve quality, but reduce the cost of quality improvement right now?

34:14Yeah.

34:15Ido Zamberg: No, I love this question because this, for the last year and a half, this was my main focus. So there's three major gaps in the world of quality. So as you mentioned, I mean, there's a lot of investment in quality improvement, but actually the investment is in extracting quality data. That's the hard part. So you have a lot of companies and a lot of internal IT efforts that goes into creating dashboards. Dashboards of saying, you know, we are, this is the percentage of compliance. The problem is, the first problem at least, is that this data is only available to a handful of people within the institution.

34:54Ido Zamberg: And in fact, it's not really available or accessible to the people who are actually providing care, so the individual provider. Because if you want to cause behavioral change, well, I need to know if I'm good or not at a specific quality standard or not. So this is the first piece. The individual clinician, individual provider, they do not have access to their individual quality data. The second piece is that, well, and unless someone can correct me if I'm wrong, but no one goes to their job as a health professional and say, well, I don't want to do my job well. If I'm below threshold for a specific metric, there's a good chance that simply I'm not sure what is the best practice.

35:32Ido Zamberg: And when you look at the definition of quality standards, it's pretty complex. Some definition can be like two, three pages long, and it's not very clear what do I need to do in order to be compliant to a metric. And then the third piece, I mean, the third gap that we found is that, well, it kind of goes back to what I said. Well, if we see that we have, so the handful of people that do have access to these dashboards, they see that there's a problem. We're not good at, like, let's say the peri-op glucose management metrics that I mentioned before. I'm going to do a root cause analysis, and then I'm going to find the problem.

36:06Ido Zamberg: I'm going to design a new best practice, and I'm going to email it to everyone, and hopefully someone will actually get it. Maybe I'll do even grand rounds or a meeting for everyone and say it verbally. I'll even go as far as hiring a person who would be there for a month and explain people physically. The problem is that this is an intervention that is done at one time point. So we'll see the compliance go up, but then it will go down again because it does not sustain over time. Actually, more than 70 % of quality interventions do not sustain over time. so what we've done here is actually solving these three kind of filling these three gaps by extracting this data providing access to the individual provider with their quality data for all the strategic metrics that that are defined by the hospital or the department in the same place in the same interface telling them well you're not good at a specific thing well here is what you need to do in order to be better so giving them access in the same place to also the best practice and also kind of a digestive of three steps that you need to do in order to be to be compliant and then on the back end we create something that is automatic and continuous meaning that people that are below threshold will continuously get notifications and emails and in app messaging from us in order to help them improve their compliance and then either you're below threshold or if you're even above threshold but you're trending down as well you'll get a reminder saying, well, you're still great, but you're trending down.

37:34Ido Zamberg: Here's a quick reminder of what you need to do in order to be better. And we've shown in multiple institutions about multiple metrics that this helps improve individual compliance. And if you improve individual compliance, well, you improve the institution's compliance as a whole. So we've shown in multiple studies. And again, if the process is better, well, the outcome should be better. Great. Well, it's wonderful to hear the overarching vision for improving quality and in standardizing healthcare and supporting healthcare providers along the way. I think we're at time. I really want to thank you, Dr.

38:10Ido Zandberg, for joining us on the podcast today. It's been a great conversation.

38:14Ido Zamberg: Thank you so much for inviting me. It's great being here.