Beamtree Holdings Limited (BMT) Earnings Call Transcript & Summary

Welcome, everybody, to this product demo demonstration of some of Beamtree's products. Grateful to everybody who's attending in person and to everybody online who's joining us from around the world. This is going to be a series of introductions to product and also some illumination of the benefits of those products from our clients and a thanks to them in advance for their willingness to provide that briefing. I'd just like to start by acknowledging the traditional owners of the lands on which we all generally meet in our various locations and pay my respects to Elders past, present and emerging. So if I could just start by moving to the learning health system, which is the mission narrative of Beamtree. So Beamtree, a business with operations now in -- or with clients in 26 countries around the world, obviously, headquartered in Sydney, but with reach into many communities has as its core offer, the learning health system. We support health systems to learn. And we do that in 3 distinct ways. We have -- we provide them with technology, which supports them having confidence in the underlying data that is being used in their health systems for making decisions. That's the first domain of the learning process and data quality. Just being absolutely certain that the data that's being used is reliable. The second element of the learning health system is that if you have good data, you can then use that for analytics, for benchmarking for comparing the quality and efficiency of services being provided. And with that comparison, you're able to identify variation in your performance and variation you'll want to reduce or increase that variation leads you towards decision support, where you want to be providing guidance, rules to your frontline teams to be able to make the best possible outcome for patients and clients that they're treating. So that's the virtuous cycle of the learning health system. If I can now layer our products over the top of that. So the first family of products, which you're going to hear about today are PICQ, RISQ, the integrated coding platform and see very early work in automated coding that we started. But the main focus of today's demonstration will be on PICQ and RISQ. PICQ, is a technology which is very well established in Australia, now operating in a number of countries overseas and PICQ, as you'll hear in a minute, provides audit of data quality in hospitals and does that extremely effectively. The integrated coding platform is a new program we're working on with initially with colleagues in Saudi Arabia and you'll hear reference to that, which provides a full service -- coding service to our hospital clients. And then automated coding is next-generation technology to actually support the assistance of coding services in hospitals. Moving from the first domain of the learning health system in a second to that area of activity, data science, comparative analytics. Here, we have 2 very important programs, Health Roundtable, by far the largest of them. Health Roundtable supports about 200 hospitals across Australia and New Zealand to compare their outcomes, and we serve them an enormous privilege to -- with the data services that help support those conversations. The ability of Roundtable is a similar collaborative of providers in the disability area in Australia. And then thirdly, we have our products in the area of decision support, the third domain of the learning health system. The first phase, of course, is RippleDown, which you'll hear about shortly. And the fourth, the Ainsoff Deterioration Index. This is the application of AI in a very important hospital context to drive improved patient safety and economic value in hospital services. So just to give you a very quick overview of how the platform the Beamtree offers fits together. And I'm now going to turn first conversely because we have a colleague from Saudi Arabia joining us a little bit later during this presentation. So we're going to start from the back of our learning health system with RippleDown. I'm going to turn to Cheryl McCullagh, our Chief Product Officer, to give an overview of RippleDown and to then demonstrate the product before leading us into a client brief. Thank you.

Thanks, Tim. Quick run through starting with a short video. [Presentation]

Let me start really quickly going through a little bit of the return on investment and why this was invented. So RippleDown Expert was invented by pathologists, for pathologists to use in the labs to make their expertise scalable and to reduce the mundane tasks they had to do. So to create efficiency for the workforce drives best practice and standardization and safety and quality can alert people to real risks but also ensures compliance with coding processes. Next slide, please. The reason it's different is laboratory information systems are very complex and very large scale. It takes a long time to get new rules into a laboratory information system. Every lab has one. But often, it can take weeks or months for you to change the process or decisions support rules in that system. RippleDown is managed by the expert in the lab, in the workflow and can be implemented and deployed in real time. That means that the person who is the expert dealing with the condition in fathom can actually change and build and iterate on those rules in real time making sufficient use of their time and making sure they don't have to do the same thing twice. The incremental build means the knowledge base gains intelligence over time and becomes even more valuable, the longer it's in place. We don't mind what system we connect with. Any laboratory information system or other clinical systems can be used and the rule building capability is infinite. Next slide, please. The expert system, which falls into the category of AI is the expert contributes to a rule set which creates the start-up knowledge base. When the cases are moved to that knowledge base they can either be automated because they are recognized by RippleDown or they may need to be identified as not fitting the established rule. They then go to a validation process or the case may be an exception to the rule because there is additional information. When those cases are validated by an expert, new rules can be added to automate those exceptions over time, creating more and more automation every time you use the system. Next slide, please. So we work in all parts of the lab from where registration start, patient registration, referrals, information management, demand management, ordering appropriate test or recommendations through to medical validation of the outputs from lab analyzers, improving the opportunity to do more tests on the same sample providing very patient-centric reports very specific to the patient's history and providing recommendations to the referral about how they should manage that patient and also improving the billing and revenue opportunities wherever they exist. Next slide, please. So these are a couple of examples. This is an example from South Africa, where RippleDown was applied to manage data entry errors because they had a varied skill in different labs across 30 different parts of the region. This is an incredible revenue optimization for them and also meant that the information went back to the correct referral. Next slide, please. This is an example from New South Wales, where one of the lab groups, public lab groups had a very high outstanding debt due to the complexity of billing processes is in health. And the implementation will be done, corrected and automated those revenue processes, reducing their debt from $7 million to $2 million in the first 12 months. But since then is now negligible. So there's automation, efficiency of workforce, but also increased accuracy in invoicing. Next slide, please. And this example is again from South Africa, where we increased the validation of the expert interpretation of results by 53%, meaning that you're really saving pathologists' time. This is a key point here. Next slide, please. And this is an example of the ROI in public labs for the -- for using [indiscernible], which is addressing the billing and revenue processes of the lab and there can be a gain of around $0.62 per episode. I mean, you think most labs are doing multi millions of them. So just an incredible ROI. Next slide, please. This is an example from one of our newer clients who has only been using the product for 12 months. And just for one test example, thyroid test they are looking at saving in hours of time of about 0.8 [indiscernible], which is significant. Certainly, an opportunity for them and having that expert available to them for other purposes. Next slide, please. This is an example, again, for most diagnosis. And I just showed you this because they are our most recent examples. Next slide, please. So I'm just going to run through a quick view of the screens. These are confusing for people who don't work in pathology but they're very familiar to pathologists where your results will be on the top right-hand side to the right will be the most recent new results will come in -- sorry, to the left will be the historical results. On the left will be the patient cases and the different categories of those and down below will be the documentation of those provided. The documentation provided from a laboratory information system is usually a simple explanation like this where it will say something like the result is either in range or it isn't and there might be a Medicare rule or billing rule around that. Just go to the next slide, please. When we add information about the history of the patient, you can get more complex in documentation that can be automated. And then you can see additional information coming in at the bottom about the patient's history or different historical values. Down on the bottom left-hand side, you can see the conditions that are contributing to that increased documentation. If you just go to the next slide, please. There are ways to introduce other conditions locally, which might be things like your referring doctor, the type of both patients in that -- whether a patient is in the hospital or in a clinic and also adding things like medication with their histories available. And RippleDown has rules that allow you to both affirm and negate different terms so that you can confirm whether that history is true or not. Next slide, please. As you add more information and you can also add integral and range changes over time. And this, again, improves the detail and patient centricity of the documentation of time. Next slide, please. Where the rules are written in RippleDown, not everyone in the lab will write rules. Some people will be selected to do that as the expertise in their field. Others will really be the validators in their field and more information can be added over time through the validation process. Next slide, please. So just -- so to explain, our reflex testing is also an additional value where if you've got a sample going through the system, reflex testing could automatically ask you to consider other test that can be added on because of the preliminary test that comes through, and this is a specific example from [indiscernible]. Next. You can also select the order validation rate to whatever the conditions you can have available. So that allows you to say, well, I might want to release 100% of normal results. I might want to consider 50% of the results in this category because I want to make sure that I'm reviewing them readily or I might say, I want to make sure that there are no automations on patients who are pregnant or neonates because they are high risk. Next slide, please. This is a work validator, which is where most people will be working where they are checking the documentation that is automated by the system. We're looking at the right-hand side, which shows the conditions that contribute to the documentation on the left-hand side is the actual documentation. You can edit this in real time. When you edit it, RippleDown don't ask you if you want to write a rule about that new edit so it closes the circle and the loop of improvement, so you're continuously improving the automation and documentation as you go. Next slide, please. We are rebuilding both the knowledge builder and the validator into one simple application. It's going to be a much more user-friendly visual. It will have sliders and click selects as opposed to click through processes. It will reduce about 20 to 30 clicks per process. It will make it much quicker to the user. And we'll also introduce our new release of machine learning enhanced rule writing that will make it quicker to the user and also more prompting. So it's an exciting year being launched next year. That's it from me. I'm going to hand over to Dr. Chris Barnes, who has joined us as one of our clients from a large group here in Australia Clinical Labs. Thanks, Chris, over to you to explain your experience.

Can you hear me online?

Yes.

Sorry. Sorry about not having the video going. I'm just running between laboratories. My role just 10 seconds of introduction. I'm the Director of Pathology with Australian Clinical Labs, where we're the third largest private pathology organization. We have 80 laboratories. I have ultimate responsibility for the clinical governance of what goes on in the laboratories and where we have had very good experience with RippleDown. It has been within our laboratory for a number of years within the biochemistry specialty, and we're adapting it to a number of different specialties at the moment, hematology and we hope to look at microbiology. But just very briefly, there's 2 reasons why I think RippleDown is -- or I can make a unqualified recommendation of this product. It solves some of our problems. Pathology is very complicated. It was listed in that presentation. When a sample comes into the laboratory, there are upwards of 20, 30, 40 different variables or different pieces of information, which can influence the result. And at the moment, we're doing that in a very manual labor-intensive heavy way which is not particularly safe, certainly not efficient. And I dare say the quality can be unreliable. And then the second reason is RippleDown will deal with those variables by me making the rules. It's -- I think to call it user-friendly, is probably underselling it a bit. I have a very good working knowledge of computers, but I don't code and yet I can create rules in RippleDown. And whereas in the past, it involved an IT intermediary, where some of the detail or some of the nuances were lost in translation, I can adjust the rules which makes it much more efficient and doesn't rely on IT, which is obviously resource constrained. So -- but for those reasons, I think RippleDown really is a excellent product in the setting of modern health care, particularly pathology, but I think you could be extended into lots of clinical domains. And I'm looking forward to certainly [Technical Difficult] Beamtree can take it.

Well, thank you very much indeed, Chris, for that. And thank you for dialing in. I know you have a busy day. So thank you very much indeed for joining us. So if I could just move on, and we're going to show you 2 -- 3 more -- well, 2 more products today. So thanks, Chris. So if I can move on now, do we have Marek online? Okay. So I'm just going to introduce the next product demonstration. So this is our other clinical decision support product, RippleDown is operating in the laboratory. As Chris says, in fact, it's got loads of opportunities to operate in other parts of the health care system. But then, we have a separate AI solution that's operating in the hospital environment, and we're moving that now and looking at ways in which we can support applications in the virtual hospital environment, which has become such an important as tactical opportunity since COVID. This is an AI application, which predicts the risk of patient deterioration well in advance of that occurring in the hospital. And it's a great pleasure for me to introduce Dr. Levi Bassin. Levi is a cardiothoracic surgeon at North Shore Hospital. He was the coinventor of this AI application. He's now senior clinical and technical adviser at Beamtree. So Levi, over to you for an introduction and demonstration of the product. Levi's then going to be followed by Dr. Jeanette Conley, who is from the Sydney Adventist Hospital, which has been pioneering application of this technology for the last 2 or 3 years. So if I could just hand over to Levi to do the presentation.

Yes. Hi, everyone. Levi is still in theater at the moment, and hopefully, he'll be able to join us shortly. Just to prove that he is a working surgeon. So I'll take you through the deck until he joins us. So if we can go to the first slide, please. So I think what's important is that lots of machine learning applications are out there. There's lots of things coming to light at the moment, AI, a very hot topic. But there's a pretty big difference between a good idea and a good algorithm and an actual product. So what we've done with the Ainsoff Deterioration Index has actually created a model, proving the model. We've published outcomes. We can show some very key features around genuine early detection, targeting clinicians with real-time alerts providing a useful clinical summary with that alert to that escalations can be made by these minimizing the number of false alerts. So keeping busy clinicians focused on the things that really matters. And additionally, monitoring acuity of all patients. So while this has been developed to detect deteriorating patients, it's equally important to know the wellness risk of all patients in the hospital. We'll talk a little bit more about that when we get to the demonstration. And the last point there around just a proven implementation is important also. So I did mention that we have proven outcomes. So here, you can see what we're trying to do is deal with this problem of misdeterioration. So despite the fact that there are early warning systems in use at the moment, we still see a fairly high percentage of people that are deteriorating significantly. And often that's because of missed or poorly recognized deterioration. And it accounts for a lot of expenditure. So you can see some pretty big numbers there on the screen around how we are still experiencing from misdeterioration. And this is WHO down the bottom, projecting a shortfall in health care workers. So not only are we having really experienced clinicians really stretched, we're also having the additional problem of less people joining the workforce and then having [indiscernible] quickly. So it's possibly not hard to understand how this problem still exists. Next slide, please. So I mentioned before there are early warning scores and they do a really good job of predicting people who are already deteriorating. So the concept of rescuing patients is well understood and will integrate it into clinical practice. But what ADI can offer is the opportunity to actually prevent that deterioration from happening in the first place. The second thing we do is really look at trends. That's highly ignored in existing early warning systems, which will measure a patient's condition against some benchmarking, some common limits, but it's really important to trend against what's normal for the individual patient, and that's what the algorithm does well. Patient variation, really, really fit people, younger people, they have different normals, if you like, and the algorithm is able to take that into account and consider variation around what's normal for you as an individual versus a common standard. I did touch on alert before a real problem. There's a lot of things going being in hospitals, and we want to make sure that our very stretched clinicians are focusing on the things that really matter. Lots and lots of algorithms out there are pretty disease specific. So they focus on one thing or one type of deterioration, example on the screen is sepsis, which is blood poisoning, and that is a very common way that people can deteriorate or deteriorate quickly. What the algorithms good at is really being homogenous across all kinds of diseases, all kinds of processes that are happening. It really just focuses on that the biometric signs of wellness regardless of what your -- you've been admitted for or what you're experiencing while you're in the hospital. Again, talking about the proof, implementing, having a product that's supportable and able to be implemented and managed is quite different from lots of really fantastic ideas that are out there. A lot of algorithms as well. They really do focus on the sickest patients, and that's kind of where we started, but what we've begun to learn about and what we can see from our product now in development is that acuity of all patients can be equally important. Next slide, please. So for the nonclinical people on the call, what are we looking for? So what AI is doing here in the Ainsoff Deterioration Index is what it is a little hard for clinicians, busy clinicians to do but pretty easy for a computer to do. And that's noticed these trends. So in the graph that you're looking at here, we can see the big purple line in the middle is where the ADI would have picked up the fact that a patient was trending in the wrong direction. But you can see it takes a lot longer for them to hit a static benchmark of yellow or red, which is wellness or unwelless. So it's really focused on looking for those trends early and giving clinicians a chance to do something about it. Next slide, please. So what we have done is taken a very, very large historical data set and used it to build a model that can produce a score of acuity for patients across the hospital. So it's a simple score from 1 to 10. If you're scoring 6 or 7, then you have moved yourself into the yellow alert level. If you're scoring 8 or above then you are in the red alert level. Level from 6 onwards, we can send alerts directly to clinicians as well, which means they don't have to go hunting for information to notice that a patient is unwell. These -- the yellow and red coloring is very familiar in the health care setting for trending in the wrong direction versus very unwell and needs immediate action. So that's the coloring that we've followed in the product. Next slide, please. So this is an image on our dashboard, which I can demonstrate to you live in a moment, which is a way for us to visually represent the wellness of all the patients in the hospital. Next slide, please. If we go into an individual patient, you can see some summary information of their vital signs and their pathology results that have been used in the model. And you can also see how they've been performing down the bottom over time. Next slide, please. So I talked a little bit about the clinical summary, and you can see here the opportunity to provide a push message to a clinician, which explains where the patients are, what's triggered their alert and provide some detailed information to the clinicians to enable escalations or to refer back to the patient chart and dig into the detail. Next slide, please. So this is our population-based dashboard. So this is across the whole of the hospital. And what this is designed to do is to show a very quick visual summary of the condition of all the patients across the hospital, drawing your attention to which wards are the sickest and which ones have the more sick patients. So you can see there on the left of the screen to north. We've got the top 3 sickest patients in that ward, but you can also see the average acuity of that ward is higher than others throughout the hospital. What this can do then is enable hospital executives to make really good decisions about where to put staff and where to put patients and basically help manage the hospital. Next slide, please. So as part of our work with the Sydney Adventist Hospital, we did undergo a 10-month clinical trial. Sydney Adventist Hospital is a 500-bed hospital. It is a private hospital, but it has a high complexity of patients, including emergency departments, all types of surgery, elective and nonelective, oncology and so on and so forth, much like you'd expect of any sort of large public 500-bed hospital. And what we found was some pretty pleasing clinical outcomes that you can see there on the screen, really high percentages of improvement in those outcomes. And then a secondary benefit, obviously, is the operational outcomes. And this is where we start to see real kind of financial metrics starting to impact organizations either in genuine financial savings or on the ability to run -- divert funds or divert effort to more patients and increase reports. So a 20% reduction in unplanned ICU admission. Some of you on the call may know that ICU beds are not inexpensive. So to be able to have that kind of impact at that ICU level in addition, obviously, to preventing patients from deteriorate to a level where they need ICU care is pretty impressive. A 5% reduction in the length of stay across the hospital as a raw percentage, might not sound like much, but even a 1% reduction in length of stay across hospitals could be representative of a multimillion dollar improvements. Next slide. So this was just an example of if you had applied a 5% reduction to length of stay, across an average hospital, then you're looking at some pretty big numbers there around savings in terms of not only dollars but also bed days, which can be used to deliver care to a greater population of patients. Next slide. So...

So thank you for that. I should just say that [indiscernible] leads have been before the Ainsoff Deterioration Index and [indiscernible] in surgery. I'm just -- has he arrived? No, he's still in surgery. Okay. So we can now turn to Dr. Jeanette Conley from the Sydney Adventist Hospital. Thank you, Dr., so much for stepping in there but very interested to hear Dr. Conley from you on really just your experience of the Ainsoff Deterioration Index .Thank you very much for joining us.

Yes. Hi, everyone. I'm the medical and clinical governance executive here at Adventist Healthcare. And I was here during the time we were doing that trial and designing our process to make sure that our ADI alerts got acted upon. It was interesting to work with our clinicians Levi and also David Bell, and refine the algorithm, refine the sorts of thresholds that we wanted to get the alerts come through it. But I think we were very pleased with the outcome, and it's now been a number of years and it's very embedded now in our process. We still -- and we -- at the time we had the between the flags type approach to picking up a deteriorating patient, the red zone and the yellow zone, when nurses went to do observations on the patient if they have blood pressure and pulse and oxygen, et cetera, we're in certain levels, then that would trigger a between the flags response. So we already have that in place. But what we have found is with the ADI, definitely, we're getting the -- we're getting alerts a lot earlier. And I feel that we've -- the way that we've set through the text messages, which [indiscernible] sort of demonstrate an example of that goes through to the team leaders, and as unit managers and also our Career Medical Officers that do the first-line medical response. It really assists them to say quickly what the potential problems are and because in the private sector, we don't have lots of teams of junior doctors running around. Our Career Medical Officers are looking after a lot of beds per doctor. The Career Medical Officer has well close to 200 beds to look after. But those text messages have been really critical because I can easily triage who do they need to go to first. And we've been really excited to see changes in the metrics that was already shared earlier. But also in our hospital standardized mortality rate, we're subscribed to Health Roundtable. And so for a number of years, we've been tracking our hospital standardized mortality ratio. And actually, traditionally, it was quite good. And then when we built a whole extra tower here and doubled our bed size, our HSM sort of crept up a bit and it got to the level before or just at the time when we were introducing the ADI, we were actually at 117, which was above the 2 standard deviations above. So we weren't where we wanted to be. And I think in our most recent report, we were down to 87. And I also got to say like I was recently doing a deep dive. Unfortunately, a patient had died, and I was reviewing that case. But it was really clear to me that every single time the ADI alert went off, nurses were taking action, they were bringing the admitting doctor or the Career Medical Officer or modifying some other things. So it's definitely been a very important part in our journey to improving some of those metrics and part of, I think, a really important safety aspect that we're really proud of here. I don't know what else you want to know? If you've got any questions?

No, no, I tell you that, that's very helpful, and thank you so much for giving us that insight. And I know you're very busy. So again, thank you for joining us today.

Yes, if that's enough, I can head off whenever you're ready.

Thank you so much. Thank you.

All right. Thank you.

Great. Thank you very much. So we may have some time at the end for questions, but perhaps today wasn't -- was just to get you as much information as we could on the products. So I'm now going to turn over -- turn to the last person, whose is going to take you through, which is the PICQ coding products, which I mentioned earlier. So I'm going to hand over to Mark McLellan, who's our Chief Operations and Finance Officer, just to give a quick introduction to the sort of suite of coding services we offer, and then he will in turn hand over to [indiscernible], who's actually here in person to give us a demo of the PICQ and related products. I would -- this then comes in 2 parts. We're going to hear from our clients of the PICQ product in Australia and New Zealand. And all being well, a great colleague, a client in Saudi Arabia at the end of the meeting. So we'll just see how that it works, but all being well, that should happen. So Mark, to you.

Thanks, Tim. I'm going to play a 2-minute video, so I don't need to explain what [indiscernible]. [Presentation]

So exits in our coding business, that's about 25% of our revenue. We don't talk -- and [indiscernible] is going to show PICQ. He is also going to show a bit of RISQ, which is another key product within our coding segment. PICQ is in about 500 hospitals, both ANZ and internationally. It is the industry standard for assessing and reporting around clinical coding data. It basically does a quality and compliance check for every clinical code. How does it do that? It does -- it uses 900 indicators against which is sort of pulled together against coding internationally recognized coding standards. It identifies errors and the source of errors. It looks at possible correction for the errors, and it measures compliance with standards and coding specificity. Why is that useful? Well, it's useful for -- it gives insight for coders, managers and executives around the activity in hospital. It allows you to benchmark the individual coders within the coding team in hospital and also benchmark clinicians and hospitals. It also allows the coders to learn on the job. So it can -- the coder can identify where they make errors. And then hopefully, they'll prevent errors going forward. It will also importantly ensure that hospitals receive the appropriate funding for activity they produce. It also provides areas for clinical documentation improvement. Next slide, please. The other project that we're going to see is the -- is RISQ, which is the relative indicator, safety and quality. So again, it uses coding data as a basis, it's connected to PICQ is focused on hospital-acquired complications, and they are extremely costly for our hospitals. Typically, if somebody goes in a hospital and they acquire a HAC, the cost of that procedure and initial procedure goes up by 5x to 8x. So it is a costly business to the hospitals. And typically, Australia will spend an additional $5 billion a year on HACs. So it's a key issue for hospitals not just in history but globally. So RISQ is a tool around HAC measurement and it's auditing tool for all HACs. The difference between RISQ and PICQ is RISQ will be used by coders but also used by clinicians. And so there's extensive workflow in these products. Effectively, we'll use -- it will review the coding records and look at the underlying data quality the condition on set flag for -- when someone enters the hospital. And that allows you to measure and benchmark all around HACs, both within hospital and against peers as well. As I quote here on the slide, just the impact of introducing RISQ into that hospital. They were focused on HACs by introducing RISQ. It helped drive a 16% reduction in overall HACs and reduced pressure interest by 70%. I'm now going to hand over to [indiscernible]. He is the Beamtree [indiscernible] on taking RISQ and he'll show a product demo.

Thanks very much, Mark. So what we're looking at here is PICQ. But before I leap into that, I just thought I'd give a quick story of what is clinical coding. So if you go to a hospital, the clinicians and nurses and specialists and some of the services at the hospital will add information to your medical record. So you'll end up -- by the time you discharge, you'll have a very big medical record with all sorts of information in there. Some of it is pretext clinical medicine literally handwritten by clinicians, some of it in a structured form such as what we saw coming through RippleDown to pathology. And so the clinical coding process is to take that complex medical record and turn it into a sequence of codes. So there'll be a sequence of codes for diagnosis, all the things that were wrong with you when you went to a hospital and a sequence of codes for interventions. So all the things they did to you while you are in hospital. So why is that important? Why do we have to capture that code of data? Well, the first reason is for research. So if we want to answer questions like how many people who've got COVID, who admitted to hospital for COVID also had heart disease. We can only get that information if we're looking at the coded medical data. Another important area is that it's standardized. So it uses a system called ICD-10, which is established by the World Health Organization. So we can do research like that across different countries. We can say what is the rate of cesarean deliveries versus natural deliveries that measure those across different countries. But the other important use of the coded data is for funding. And this is an area where Australia has really been a pioneer. So Australia for 20 years or more has been using a system called ICD-10-AM, which is the Australian modification of ICD-10, that includes a lot more detail around the codes and it enables what we call activity-based funding. So this is a way of funding hospitals by grouping episode to hospital care into categories and then essentially putting a price on that category. So there's some 600 or 700 possible categories, which we call a DRG or a diagnosis-related group and Medicare or in the case of private health care, the health funds will literally put a price against each of those adjusted for complexity, so it could be the patient's elderly or they had to stay in hospital for a long time or could also be adjusted for the incidence of hospital-acquired complications. So this is a recent -- relatively recent change to the way that funding works and the hospitals are now starting to be penalized for avoidable hospital-acquired complication. So again, that level of penalty is determined from the coded data, whether it's Medicare or the health funds. So there's a lot of people very interested in making sure that the coded data is correct, and that is where PICQ comes in. So PICQ has been in operation for a long time in Australia. But in fact, because activity-based funding has proven to be such a successful model for [indiscernible] in health care costs, the Australian system has been adopted in other countries around the world. So Ireland, Singapore, Saudi Arabia, long tail of Eastern European countries. And of course, these are the countries where we are selling our products, which have been proven in Australia but now adopted in Ireland and some of these other countries. So what we're looking at here is PICQ. PICQ consists of workflow tool for clinical coders. So the people within the hospital that perform the coding function, they have a workflow, which allows them to work through areas that have been raised by PICQ either because the coding is recognized as not being correct. So not being compliant with standards or because it lacks specificity. So making sure that you get the specific code as opposed to a more general code is also very important for making sure that you get the correct DRG. So compliance and specificity. And typically, a clinical code will land on this dashboard. Every morning, they will go through the coding they did the previous day. They might get 2 or 3 PICQ areas that they need to correct and then they will go in and correct them and/or leave some commentary. I'll talk a little bit about that in a moment. The other aspect PICQ is a reporting capability. So managers, health information managers who are responsible for the coding function within the hospital as well as health service executives who are interested in the outcome of the coded data can view reports, which allow them to measure the coding quality, both within their organization as well as benchmarking that against -- anonymously against other hospitals that are users of PICQ. So they can determine not just an arbitrary value of coded quality, but it is better than my peers versus MI and outliers [indiscernible] my coding quality provider. So what we're seeing on the screen here is the dashboard. We -- PICQ uses the concept of indicators, as Mark said. And those indicators are like rules that are triggered because they identify a particular coding problem. It's a standard traffic-light approach. The more serious areas are in red. Most clinical coders will -- at least the red and orange areas, and if they've got time, they'll go back and review the yellows as more looking at it in terms of trends rather than correcting every hour. From a management point of view, I can also see the trend in those over time. So I can see whether or not my teams' clinical coding is maintaining at a certain level or declining or improving. And I can also see what types of problems my team is having. So what we're looking at here is every possible disease chapter so when these clinical codes, they come in chapters related to different body systems. So I can identify, for example, that here, this particular team seems to have a problem with coding injuries. And that's not a real unusual actually injury coding is very complex. And it involves coding not just the -- what was the injury, but how did it occur? When did it occur? What was -- what activity was the patient undertaking when the injury occurred? And so using this information, the health information managers can drive professional development. Today, coding teams -- coders can also manage their own ongoing professional development. And this information can then be used also for driving clinical documentation improvement because, of course, if the clinical documentation is inadequate the coders can't adequately abstract the correct codes from that information. If I run a service that's got many hospitals, I can also then compare those hospitals against one another for coding quality. I can also -- because the way the PICQ works is that we're continuously receiving a feed of data from all of the hospitals around Australia. We can also see when the coding has been changed to correct the error. So I can see not only what was the initial coding error rate, but also after correction, what was the final error rate. Of course, what we would hope to see is that after correction everything has come down to 0, but in practice, that takes time. We actually have customized views within PICQ for public and private hospital workflows because they are quite different and the way in which the funding is allocated and the time lines around having correct coding completed are very different in the private sector versus the public sector. So we provide different ways of peeling that information. In addition to being able to -- also what I might just do is give you a quick example of an episode that comes into PICQ. So here is an example of the patient devastated after that patient has been discharged and the medical record has been coded. It will come into PICQ as a series of diagnosis codes. So we can see here this particular patient looks like they have been admitted because they had a complication with a artificial hip joint. So these are all the diagnosis codes. You mentioned that cough condition onset flag earlier. So this is a really important piece of information that says did this diagnosis present on admission or did it occur while the patient was in hospital. So as an example here, we can see that they -- when they were admitted, they had this mechanical complication of the hip joint but actually, while they were in hospital, there was hospital -- avoidable-hospital complication which was damaged the cartilage during the procedure, and there's some additional information about that in the record as well. We can see the procedure codes. What did the hospital do to the patient while they were there. And down here, we can see this has triggered one PICQ indicator. And I can see that this is an indicator of degree F, which is the most important degree and I can see that it is not yet been fixed. There's some information here, which is the rationale, so I can -- that's a clinical coder I can -- I won't expect you to read that now but that's information that's useful to the clinical coder that says, why has PICQ alerted on this and what should I do to fix it. So the coder will then go in and say, okay, I can see Mr. Code. They will correct the coding and then they'll mark that it's fixed. And so in that way, they will work through their -- they work with. People also check to make sure that it has been fixed. So when we receive a new version of the episode, we can check that and make sure that that's right. Now sometimes it's the case that for certain types of indicators, there is a small false positive rate. So there's a small number of conditions where actually perhaps the coding is correct as it stands. Those specific conditions are noted in the indicator and often that will be something like if the patient is a pediatric patient or the example that I often give you is, where there's a particular indicator that looks at where a pacemaker has been inserted and it wants to make sure that you consider not only the pacemaker, but also the leads. So the coding for that is done. But in juvenile patients, that can be done as to separate episodes. And so that's the small number of cases where that would be considered -- the coding would be considered correct. So if that's the case, the coder can mark it justified. They'll then select from a number of reasons. Why is this justified? And those are related to the specifics of the indicator. And that information then goes back to the coding manager. So there's some additional workflow there around justification. Now there's a second role that the coder performs when the hospital is using both PICQ and RISQ. So PICQ for managing the coding quality, RISQ for managing the hospital-acquired complication rates. And that additional step is that the technical coders also received information about the HACs and they will then go and confirm that the medical record truly reflects the fact that it was a hospital-acquired communication. And that's very important because as I mentioned earlier, the hospital is actually penalized for HACs that are considered to have been avoidable. So going back to that second check, it's very important to make sure that the coding is correct. I'll just talk a little bit more about some other reports that are available. So one of the other things that we do within PICQ is that we look at the DRG. So again, the DRG, the diagnosis-related group, this is like the price, if you like, of the episode. The amount that the hospital will be reimbursed either by the government or the health fund for that particular episode of hospital fee. And what we can track is how did this episode change over time from the first time it was coded until the final time that was submitted. And that's important because if there is a consistent lack of attribution of the correct DRG, then that points to the fact that in the general sense, this hospital has not been properly reimbursed for the health care it's providing. So one of the things that we can analyze here is the total number of times that the DRG changed in response to coders going back and reviewing the episodes. And of those, how many was because of complexity? Well, how many were because the actual DRG was incorrect? So by and large, it is a complexity issue that the coder may not on the first instance, have picked up certain comorbidities or additional diagnosis, which would cause that DRG to be bumped up to another bit of complexity for which the hospital would receive additional funding. So this is very important to hospital health information managers and health service executives to track the DRG changing to make sure that the hospitals are getting the correct DRGs through their coding. Another aspect is where we provide hospitals with the ability to benchmark themselves anonymously against all other users of PICQ. So we group hospitals together according to the profile of patients that they see. So the case mix, a number of other variables that we used to establish who are the statistical peers of your hospital. And then health information managers have access to a report that allows them to view how their quality ratio is tracking against others. So they can see not only what is the absolute quality ratio that they're seeing, but how am I tracking against hospitals similar to mine with similar sized coding team seeing a similar number of episodes at the same time. And that's very valuable, particularly when new rules are introduced. So we saw a lot of activity around this during COVID because there were a number of COVID specific rules that hospitals were required to use when they're coding. And so this was very useful to hospitals in terms of being able to track that initially a spike in coding quality problems and then seeing that decline satisfactorily over time. So I'll just briefly switch over to talk about RISQ. So RISQ is deeply integrated with PICQ. It uses exactly the same data set exactly the same coded medical records. But in this case, instead of looking at the coding quality, it's looking at the incidences of hospital-acquired complications. These tool is used more by [ craft groups ], clinical governance teams rather than the clinical coding team directly where they are looking -- they may want to go in and look at all the hospitals in their network, and then drill down into that to see which particular types of hospital-acquired complications are we seeing problems with? So here, I can see this hospital, which is just a dummy data, obviously. But this hospital is equal to peers, whereas we seem to be better than peers across some of the other hospitals. So why is this one not tracking as well. So I can drill into that and I can see that I'm health care associated reflection. I'm doing better than my peers, but for some of these other surgical respiratory and so forth, I'm doing worse than my peers. And so the concept of peers is very important here, where we do a form of RISQ adjustment by analyzing the statistical data of the hospitals to find anonymous peers that we can benchmark this hospital again. So again, similar to the quality ratio report where we are determining whether or not this hospital is an outlier. And of course, all hospitals want to be best practice, but if this provides them with an objective measure of where they fall. So this is, again, integrated with the PICQ data quality workflow because one of the interesting things that we often find is that it may look, for example, as though I have a better than peers rate at this hospital across all HAC indicators. But actually, the data quality of it is worst than peers. So what that suggests to me is if I perform some improvements to the data quality using PICQ, then what I should also see as a more accurate HAC rate. And again, because the HAC rate speaks directly to penalties in funding, it gets a lot of attention at low levels within the organization and I think that's probably enough for me.

Thank you, Mike. I think I'm going to hear from, I guess, [ Neelam Aryan ].

[ Neelam ] is a clinical coding coordinator for the [indiscernible] Health District. They uses PICQ so hoping to hear from [ Neelam ]. Hopefully, she is online to tell us about the use of PICQ and how it helps them in their objective. Over to you, [ Neelam ].

Just let me know if you can hear me.

Yes, we can. Thank you.

Great. I was obviously asked by Doug to speak about our experience with the PICQ. So here, it's [indiscernible] District. PICQ was basically implemented back in June 2017. And there are about 21 coders across the LHD that use the product like we pretty much review our episodes using PICQ. And I just wanted to share that on average, we review about 647 episodes in Affinia and the impact that we had for this financial year, Affinia '22/'23, there was about 106 episodes that actually had a DRG change just purely by using the PICQ tool. And now as Michael already said previously that one of the major reasons we have this is because we want to make sure that the DRGs that are assigned for our episodes are correct, and it reflects the correct patient complexity. Our PICQ and, of course, our coded data is used for benchmarking and it's health service planning, and like I said, the major users of the coded data is, of course, to research hematological studies helping us with public health strengths and it's, of course, impacting on our cost and budget allocation. Yes. So pretty much, we've been with the PICQ from 2017. We have no issues with them. They've been really great with us. And like I said, this is just one of the coding tools that we use to ensure that our coded data is, of course, accurate and we are getting the correct funding for the episodes in conjunction with the other audits that we do. Now with PICQ, this is something that we wouldn't be picking this up in a normal audit. We do go through another audits like DRG audits, coder audits and length of stay audits but PICQ, basically, the tool helps our coders pretty much just helps them on a daily basis of certain conventions that we manage on a day. And an example I would give just quickly, is if a coder assigns division of a -- division C, if they assign a IT code during an episode and they forget to assign the actual diagnosis code, which is for a addition for a female -- for a peritoneal addition. So that will be like a trigger for them to check the next day. It's a fatal error, right, that you've assigned IT code but you have actually missed a diagnosis code. Now this is something that wouldn't be picked up in a normal audit. So for that scenario, definitely, the PICQ tool is assisting us to pick up on certain Australian coding standards, if they are not applied correctly during the day while the coder is coding. That's all I had. Did anyone have any questions? I had only 5 minutes.

[ Neelam ], thank you so much. Very kind with your insights. So again, I know you're busy. So thank you very much for joining us.

Thank you.

Thank you. Thank you so much. Well, look, so we come to the end with a very important -- so we come to the end of the presentation, but with one very important remaining item so we recently announced a new partnership, a new product partnership with our partners in Saudi Arabia, Lean Business Services. For those of you who have been following our progress. Lean is a government company in Saudi Arabia that is leading the field on technology innovation in health care, and we are very privileged to be working with them. And we are surely be joined by Fahad Alsaawi and by my colleague Marek Stepniak, who is the Chief Product Officer from Beamtree. She's actually in Riyadh at the moment. And Marek is going to introduce a little bit more relationship with Lean and indeed the focus we've currently got on taking the PICQ product you've just seen and building that into a very exciting new platform which we are probably going to make integrated coding platform, which will offer a total transformational service to health care globally, starting in Saudi Arabia, supporting them, not just with the audit functionality of PICQ as you just heard, but also much more broadly with really solving many of the problems across the whole of this coding data quality pathway altogether. So I'm going to hand over to Marek Stepniak and then for him to introduce Fahad Alsaawi to us. Marek, you're on mute.

Good morning from a very nice Riyadh morning. Look, we're in -- sitting in the offices of Lean, and I'm with, as you heard with Fahad Alsaawi, who is the Chief Data Officer and Deputy CEO of Lean. We have set up a partnership with Lean. We're underway working both on product development but also market development. And Fahad has very kindly agreed to share a little bit more about the Saudi health system, the reforms that are taking place in that, how that impacts the relationship that we have, the opportunities that it brings. Our go-to-market partnership both locally on coding and the opportunities beyond Saudi. And Fahad, perhaps I'll just hand over for you to share a little bit more.

Thank you, Marek, and hi, everyone. I'm sure that you are aware and you heard about the transformation that is happening in Saudi. And one of the main transformation that is happening is on health care. So currently, our funding of health care ecosystem is based on budget that is coming directly from permit. However, in the coming 2 to 3 years, the whole thing will be transformed into capitation and value-based health care where data and quality of data and coding is very essential in order to do the costing and then the funding. Therefore, we started working with Beamtree on auditing, checking the quality of data and quality of coding for our hospitals and HISS. And we noticed that there is a huge gap when it comes to quality of coding audit RCM practice and also how to cost the encounters and the activities within the hospitals. And we decided with our partners at Beamtree that there is a huge gap in Saudi and also outside Saudi. We decided to partner with Beamtree and developing a holistic and comprehensive coding platform that includes not only auditing rather than code finding, DRG, costing and also RCM activity. And we signed this agreement a few months ago, and we are very eager and interested in expanding and scaling in Saudi. The market is huge in Saudi and we already started some of the engagement in Saudi. Our next step is to go into Gulf region where the same thing is happening. So hopefully, soon [Foreign Language] we will go beyond Saudi. And then we will expand internationally, even beyond the Gulf countries.

Thank you, Fahad. We're excited and privileged to be able to work with Lean on what is going to be a very interesting but also for us, both an exciting future in Saudi, Gulf and beyond that. Now back to you, Tim.

Thank you so much. Thank you, Fahad. Thank you, Marek. I know it's only there. So thank you very much for joining us. And behind them, you can see the external decoration what's called the Digital City, which is an amazing spot in which Lean is located, which is a real hub of innovation. So thank you very much indeed for joining us. So with that, I bring to a conclusion the demo. This will -- has obviously been recorded, so we can make sure that others are able to see. But thank you for joining us both on the line and in person today. Thank you very much.