Welcome, everyone. My name is Ritu Thaman. I'm an associate professor of cardiology at the University of Pittsburgh. And I will be your moderator along with judging. And we have two of our very illustrious judges here on the platform, Dr. Mintu Thakaria, and also David Kim, who you need no further introduction. The rules of the game, and by the way, all of these companies have won by just being here making it to this stage. So the rules are as follows. We will give each company 10 minutes to present their ideas, after which there will be five minutes of questions from the judges, from anyone in the audience. I will moderate those questions. And to keep the game fair, everyone will have a timer. The timer is right here. So they can all see how much time is left. And without any further ado, I would like to start with the first presentation, which will be from M. Palo. Can everyone hear me? Good morning, everyone. I'm Ray, co-founder and CTO of M. Palo. We're an MIT spin-off that brings AI-powered virtual clinic to heart failure management. Heart failure is a leading cause of hospitalization and death in the United States. Every year about a million hospital admissions due to heart failure happens in the United States. The American health care system spends over 30 billion dollars on heart failure, yet half of those patients die within four years of diagnosis. Evidence has shown that many of those patients could live for decades if they were put on optimal therapies. Guideline-directed medical therapy or GDMT has substantial benefits. It reduces mortality by 70 percent, reduces hospitalizations by 30 percent, and improves quality of life for those patients. Traditionally, a patient needs to be seen by a doctor every other week from three to six months to be titrated to maximally tolerated dose of GDMT. In the current health system, due to limited capacity, only less than 10 percent of those patients achieve optimal GDMT doses. On the other hand, hospitals have strong incentive to reduce 30-day readmissions. 30-day readmissions are not reimbursed. High readmission rates lead to reduced reimbursement from CMS and lowered star ratings. We address those pain points and generate incremental revenues for our partner health systems. We transform heart failure patient journey by shifting care from expensive hospital setting to the more cost-effective and convenient home environment. This is the status quo. A heart failure patient gets decompensated, admitted to the hospital. After several days in the hospital, they're discharged. At home, 20 percent of those patients within 30 days exacerbate and readmitted to the emergency room and admitted to the hospital. With our AI-powered virtual care, we help our partner health system to treat medications and optimal therapy in the hospital to shorten hospital stay and determine discharge readiness. When a patient is discharged to the home environment, we offer high-touch virtual care to those patients to prevent exacerbation and manage their fluid status. We enable remote nurses to offer at-home virtual care to heart failure patients, powered by our AI and under cardiologist supervision. We deliver heart failure care cost-effectively. Traditionally, medication titration is done by cardiologists and advanced practice providers or APPs. With our proprietary AI platform, we shift medication titration to nurses and software solutions. We have built co-pilot AI for medication titration and risk stratifies patients to help nurses escalate care if needed. From the patient side, with improved capacity, they'll be able to see our providers and have virtual titration appointments every other week to help them achieve optimal GDMT therapy. Our AI platform reduces trial and errors and reduces cost. Our patients can enjoy improved experience and extended quality of life. Our proprietary AI models are trained on millions of patient records. As our virtual care grows, we plan to integrate data from wearable devices, implantable devices, as well as patient self-reported data into our AI system. Here's how it works. A provider logs into our AI platform and they can see who among the populations they are monitoring are at high risk of exacerbating. For high-risk patients, a nurse can escalate the care to cardiologists or advanced practice practitioners. For those stable patients, they would remain with nurses. If a medication titration or therapy planning is needed, they can enter a hypothetical treatment regimen. In this case, for patient Gus, a provider might wonder how this patient biomarkers would trend in the next 24 hours if they give 100 milligrams of IV diuretics, of IV frosamine. And then they can see in the next 24 hours, our AI model anticipates this patient's creatinine would go up by 0.2 milligrams per deciliter. They can optimize and adjust their therapy based on our AI model's predictions. We are working with three hospitals, have three active hospital pilots, and more than five additional hospitals in the pipeline. We work with the US VA health system and MaineHealth. The market is large. We start with our hospital partners and expand to direct payer contract. We're addressing a total market of $14.4 billion. The market is also growing. We see even more upside as care shifts from fee-for-service to value-based care. The projection here is by 2027, about 130 to 160 million lives in the US would be under value-based care. We are a strong team. Our leadership has deep experience in healthcare technology and cardiovascular care. I'm Ray, co-founder and CTO of Empalo. I earned my computer science PhD at MIT. That's where I met my co-founder Claire, who had extensive business background in finance and management consulting. Dr. Dana Weishar, our director of clinical affairs, has used to lead heart transplant services at Kaiser Northern California. If your health system is interested in improving patient outcomes and free up more of your cardiologist time, please reach out to us. I'm excited to talk to you. Thank you. Thank you very much. We will now take some questions. One of the first ones that I have is, what are the data from the patient side that are being entered into your algorithm besides just blood tests? Are you incorporating other sensor data or other clinical variables? Will that algorithm be iterated as you acquire more data? Then the second part is, they're doing, I'm sure you know of this, the NHS in England has a similar model where they're doing acute care at home and they're incorporating many different variables and they've shown good outcomes. One is the strong study, if you want to look it up. But anyways, if you could answer that first question, then we'll have some others. Thank you. For your first question, right now the data that goes into our AI model includes not only lab tests but also the patient's demographic information, their comorbidity, their vital signs. And as you mentioned, while we collect more patients' data as well as data from different sources, such as wearable devices, implantable devices, or even patient self-reported data, the model can continuously learn from those trends and improve. Thanks for the presentation. There are a lot of AI heart failure companies out there, so can you explain how you're different, how you're gonna win, what your advantage is? And what I don't have a good sense of is your business model. When we compare ourselves to other AI heart failure companies, our differentiation is we also provide care. We provide virtual care powered by our AI, but we do this more cost-effectively because with our AI, we enable mid-level practitioners, such as nurses, or sometimes we can even use software solutions to help our patients titrate medications. And the second part of your question is a business model. How it works is we work with our partner health systems. They refer patients to us for outpatient medication titration. We help the system to manage those patients in the outpatient and home environment. And when care escalation is needed, for example, if we need an imaging test or there's some procedures that need to be done, we refer those patients back to the hospitals. The hospitals can either bill on behalf of us and pay us a fixed amount of fee, or we could directly bill to insurers. So those would be the two ways we can generate revenues. Sorry, I can't hear you. For the AI part of this, do you feel like you will need to go through the FDA to get that cleared for the FDA? Because part of it is you're using it for your own services. I'm kind of somewhere in the gray zone, but, you know, for many of the other technologies that are out there, they feel like they need to have a regulatory strategy to validate their AI. So where are you with that? We do have a regulatory strategy. For our internal use, as you mentioned, we actually don't need to go to the FDA. This could be considered as software as a medical device and investigational device exemption, as long as we can prove that. If we want to license our technology to external clinical users, that's when we potentially would need to get 5-10k clearance. So the next one, a follow-up question, is for IV medications, you talk about diuretics, you know, furosemide. I'm wondering, how are you going to actually execute that and make it also economic to actually deliver IV medication at home? Because that's it now, you're taking it to a different level. It's easy enough to say, hey, you know, you may need to increase your, you know, PO meds in certain ways, but to do IV meds, now you're taking it to another level. How do you make that even economical? Thank you for giving me this opportunity to clarify. That example I gave was actually for the inpatient environment, but for our AI platform, we can also, for example, look, if you give oral diuretics or furosemide, 200 milligrams, what's going to happen to this patient's biomarker trend in the next 24 hours? Thank you very much. We will move on to our next contestant. Thank you. NucleusRx. Good morning. Can you all hear me okay? Hi, my name is Projit Awan. I'm the co-founder and CEO of NucleusRx. At NucleusRx, we have built a virtual care management and medication therapy management platform, focusing on heart failure patients primarily. This is my mom. Here, she was arranging my dad's weekly medication several years ago. My dad was a cardiac patient with many other comorbidities, and they struggled to keep him compliant because of the complexity of medication management. He was taking about 18 different pills at this time. Around the same time, I was building Verizon's IoT platform, Internet of Things platform, ThinkSpace, and eventually I scaled it up to 24 million devices. I've seen many industry use cases, but yet to come across one in healthcare that could simplify such management on a daily basis. What I mean by that is having the providers connect with the patient or care provider helping the patients to manage it, easy way. That's how NucleusRx was born. Eventually, my dad passed away during COVID because of his underlying health conditions went unnoticed for a long time. He couldn't see his doctor. That gave us further food for thought that it's not just medication management. There has to be a holistic virtual care platform that can help patients like my dad. Heart failure is one of those diseases, and my fellow presenter, Ray, has spoken deeply about it. We see two issues. First of all, the management of heart failure is inadequate and it's very costly. The numbers speak for themselves. There are two issues of the heart. One is post-acute implementation of GDMT that, again, Ray spoke about is poor. And second, longitudinal care is also inadequate. Fluorosamide, for example, fluid volume overload is an issue. It is one of the causes for frequent hospitalization, but it can be prevented at home. The way we are addressing this problem is by our system. Let me just explain further. Our virtual care platform, our vision of virtual care platform or virtual care is done in three steps. One is manage patients home well, at home. Second, monitor them effectively. And third, intervene as needed. And it has to be real-time. And this is what we are building or we have built at NucleusRx. First, managing patient home is not about just medication management, making sure they're adherent to their medication, but also their diet, lifestyle, including patient education, right? It's very important for them to be educated about their disease condition. And we have looked at the reasons for this implementation failure currently in the system. There are three areas, patient-centric, provider education, and care provider issues, right? So we are addressing the first and the third, not provider issue, provider education, but patient management at home and care providers tool to make it easier for them to manage patients. That brings me to the second point, which is monitoring patient. Monitoring patient, a lot has been talked about even yesterday, data overload by remote monitoring system. Yes, we collect data from patients, their physiologic data, their qualitative data through our monitoring app. What we don't do is inundate the health care provider with data. So our machine learning algorithm is done to make it more exception-based monitoring. And third, intervention. So yes, we can detect a situation when a fluorosomide dosage needed to be updated, but how do we do that, right? Or GDMT implementation. Yes, we have talked about GDMT implementation, but how do we make the horse go to the water, right? And this is what we are doing. What we are doing is allowing or empowering the care providers and the providers to update doses remotely and in real time using our system. So this is how we do it. At the center, at the core of our platform is a pill dispenser. It's a patented connected pill dispenser. And the pill dispenser is not just another standalone pill dispenser. It is data-driven and is connected to providers EHR system. We read the e-prescription directly from the EHR and program the pill dispenser. This is how we create the connection, right? One-to-one connection between the provider and the patient. For management of the patient, we have provided physiologic devices, physiologic monitoring devices such as a weight scale and blood pressure machine and the companion app. Now we collect the data. And as I mentioned before, we only alert the care provider when the exception is met. And they have a dashboard to actually prioritize patients. And that also improves their daily operations. And so that they can focus on where their time is most valuable. Looking at the most critical patient of the hour. Instead of calling patients on a round-robin basis to make sure they are doing okay. All right? And the third is intervention. When they need to update the dose, fluorosemide dose, to prevent that emergency situation or episode, they can update the dose remotely. And it will be impacted in our system in a real-time basis. So that next dose dispense would be updated. The same way we can monitor and improve on GDMT implementation on a real-time basis and on a frequent basis without having to bring the patient to the outpatient. Heart failure virtual care monitoring market is about 25 billion dollars. That includes invasive devices, non-invasive devices, services and other things. Our focus is stage C and about patients. And that market is defined to be about $3.5 billion. And our serviceable optimal market is $260 million. There are other players that are focusing on primarily remote care or remote monitoring data to gather the data and provide services around it. Where we differ and value add is really the medication management on a daily basis. And not just that, it's updating the medication doses when it is needed the most. Again, implementation is our focus. And other players in the market fall short right there. And this is where we are providing the technology to be able to grease the path to make the implementation better. My background is technology. My co-founder, Ashok, has a deep clinical or healthcare delivery background. We are backed by a strong group of advisors, including two world-class heart failure doctors. We're currently working on a pilot study with one of the Tier 1 healthcare systems. We recently awarded a grant from NIH. It's a Phase 1 grant to support the study. We are also working with other Tier 1 healthcare systems to run further pilots. Our product is IRB Ready. We've been granted a patent, and we are working on other issuance. And our regulatory strategy is complete. In the past, we have participated in NSF's pre-accelerator program. Last year, we completed the medical device focus M2D2 program run by UMass Lowell. And last year, we also made it into the semifinals of Boston Scientific Connected Patient Challenge. Currently, we're raising $1.5 million to accelerate our commercialization. To recap, we are playing within a large market opportunity with significant unmet needs. We're a platform technology with heart failure as our beachhead focus. That means our platform can serve other disease verticals and scale-up opportunities. We're a strong team, balanced team, with strong execution experience. And the cherry on top here is that we have received a NIH grant. The success of that, we are very confident that we are going to get more non-dilutive grant money. Thank you. Thank you very much for your presentation. Very quickly, one first question is, your dispenser, it alerts the caregiver if a dose is missed, right? But what about the fact that 30% or less of our heart failure patients are not on GMTP, as we know? And how do you know when to escalate which drug? Is that going to be part of this therapy or this device? First answer to the first question is yes. So we track adherence based on two factors. One is when patient or the caregiver to the patient, authorized caregiver, they dispense, they actually acknowledge the dispense, right? And second is when they take the cup after the dispense. So those two are indicators of whether they are cognitive of their dispense time. And second, whether they're taking the cup to actually consume the medication. That data is recorded and we actually manage pill-by-pill or medication-by-medication adherence or aggregate adherence. Second part to the question is, yes, not everybody is on GDMT, but I talked about two use cases. One is GDMT optimization and the second is longitudinal care for existing patient. Maybe they have gone through post-acute care, GDMT optimization happened or not happened, but they're on a five, seven, eight-year journey with their health. Our system can benefit them as well by, like my dad's case, 18 different daily meds. When I speak to our advisors, 12 meds are normal. So yes, that's a huge cumbersome process on a daily basis. So we are helping that primarily. Another question about regulatory. Since you're going to be adjusting medications, it sounds like the software is going to be at a minimum recommending it. I'm just wondering what is your thought about what kind of regulatory validation or regulatory approval you'll need to do that and what kind of clinical validation you're going to require in order to allow that process to go? So regulatory validation, that's actually easier because we have pretty good devices in the market. There are stand-alone pill dispensers available. What we are doing is we are adding SAMD, software as a medical devices. What we are not getting into as of yet is a clinical decision system. So that's how we are keeping it regulatory-wise easy. So it's going to be acknowledge acceptance. For example, if the doctor changes dosage from their EHR, it has to be acknowledged by the patient to accept it. But then you just remove the real-time effectiveness of that because getting information to the doctor and having them act on it may not be real-time, even if you connect the other parts of the communication process. So how do you make it real-time then? So our goal is to make it real-time eventually. I think what we want to do is to commercialize. And then, again, we are not working as an independent system, stand-alone technology. We are working with the healthcare providers that are actually purviewing. The patients are under their purview. So when they make changes, the goal is to have them directly call the patient and make sure they accept it. Yeah? Very. I think we're running out of time, so just a quick comment. And then if you can answer a question quickly. The hard part on the comment is anything that requires EHR integration for customer acquisition is insanely hard, and many companies have failed. And so I would encourage you to rethink that as part of this. The question I have, if you can answer briefly, is what is the single outcome that you're solving for? Are you solving for mortality? Are you solving for time to optimal GDMP? Are you solving for labor costs? I would say all of the above. What's the number one? The number one is improvement of mortality. Thank you. We will move on to our third pitch from ... Can everybody hear me? Okay, good. So when I was putting this presentation together ... Oh, this is CardioSense, CardioSense? We will start now. Okay, great. Thank you. Hopefully everybody can hear me. Good morning, and thank you for inviting us to be part of this pitch competition. I'm Emery Nan. I'm the co-founder and chief scientific officer for CardioSense. I'm also a chaired professor of electrical and computer engineering at Georgia Tech, which made it easy for me to get here in the morning, so welcome to Atlanta. I want to tell you a little bit about what we're doing at CardioSense. So we're working on non-invasive wearable sensing devices and associated algorithms, deep learning algorithms, to really enable early detection and personalized guideline directed care of patients with heart disease. I don't have to go into these numbers. Thank you for the presenters before me, but basically the first problem we're going to talk about is heart failure management, and specifically what we're going to talk about from the CardioSense standpoint is the use of filling pressures, which with implantable technologies like CardioMEMS and Edotronics have already been shown in large clinical trials to be very effective at reducing mortality and reducing hospitalizations in patients with heart failure. In fact, when we talk about GDMT, really what we're talking about is primarily being able to use pressures and the volume status of the person with heart failure to then follow a guideline directed process for up titrating diuretics and other drugs to be able to manage their care. The problem is that implantable devices like CardioMEMS and Edotronics that are available both FDA approved for heart failure are only available to a small fraction of the patients that have heart failure. So less than 2% of all patients with heart failure have had such implants. Even though these implants are very effective at allowing for titration of drugs and reduction of hospitalizations, reduction of re-hospitalizations, they're just not widely available. So what we're working on is a non-invasive sensor with deep learning algorithms that extracts information from the chest that can be used in clinical settings, outside of clinical settings to be able to derive pulmonary capillary wedge pressure estimates for patients with heart failure. And specifically our wearable device, which looks a lot like many of the Holter monitor patches that you see around here, actually measures three modalities of waveform. So it measures the electrocardiogram, of course, and optical measurement of the PPG, which is what you use for SpO2 and on smartwatches. But also we have a very small low noise accelerometer to measure the tiny vibrations of the chest wall in response to the heartbeat. And this is a signal called the seismic cardiogram that many people have never heard of, even fewer have ever studied it, but it happens to be something that I've studied and are other key technical leads in cardio sense have studied for more than a decade at this point. And it turns out to be a very important signal because ultimately when we talk about filling pressures, you're talking about mechanical information from the body. And so that information is not really encoded in the electrical signal of the ECG, nor is it encoded in a peripheral vascular pulse signal that you get from the PPG, but it does happen to be encoded in these mechanical vibrations of the heart that can be measured on the sternum But what's more important to what we're doing is that we have deep learning approaches that can take waveforms in and produce waveforms out. And this ends up making it so that with the patient data that we do have, and I'll talk about that in a moment, we're able to build very powerful models that can accurately estimate on an absolute level, the wedge pressure for a person, just with this noninvasive data at the same level of accuracy that you would normally get with an implantable device. This is exciting to us. So where we are right now is we've completed a multi-center, 15-site study of 1,000 patients with heart failure, as well as some suspected with heart failure, where we had our device simultaneously with right heart catheterization, and we were able to measure the waveforms from our device and the right heart cath synchronized from all these patients, of which 350 had HFREF, or heart failure with reduced ejection fraction, which is what we're focusing on first. And we had great diversity among these patients in terms of their demographics, in terms of their other comorbidities, mitral valve regurge, atrial fibrillation, and all the characteristics that you'd expect. And with this, we were able to produce an algorithm that has, again, an accuracy that's kind of on par in terms of error with respect to cath as compared to these implantable devices that have already been approved. And I'll show you some of the waveforms here, because for me, being an electrical engineer, actually, I like seeing waveforms, and I think actually this being sort of an arrhythmia and EP-related conference, I think that people in the audience hopefully will as well. So with our device, we have, as I mentioned, three different modalities of waveforms. The ECG and PPG probably look familiar to many of you. The PPG we're measuring reflectively off the chest. The third signal is the size of a cardiogram, which has some information, especially early on in systole for the sort of main thump associated with cardiac ejection, and then also has some information late in diastole that ends up being actually very important to what we're trying to do here. And then you can see here, basically, that those signals are measured for this person at the same time as right heart cath in multiple different chambers, so including the pulmonary capillary wedge pressure, but also pulmonary artery pressure and right atrial pressure as well. These are now the waveforms from two different patients. On the left side, you have someone with normal sinus rhythm that has ECG, SCG, and PPG from our device. At the bottom, you see a right heart cath that's also measured at the same time, this waveform, in green, and in blue is the output of our model, and of course, this is one of the people where we have almost no error in the right heart cath estimation, and I should mention that all of the right heart cath waveforms go to a third-party adjudicator site, in this case, Duke, where they have two or three different adjudicators look at the waveform and output the actual value of PCWP. This turns out to be very important because the machine output of PCWP value has quite a bit of error compared to what adjudicator provides. And then on the right side, actually, we selected someone with AFib, just to give you an example of where you can have heart rhythm abnormality in this case, but at the same time, you have the right heart cath waveform. Of course, there are some of these waves that are associated with AFib that are not predicted, but the average right heart cath PCWP mean value is actually predicted very accurately in that case as well. So what we're trying to do is we're trying to provide a tool that can be coupled with the existing GDMT approaches to improve the management of patients with heart failure, with hemodynamic and GDMT optimization. The idea is that we would have our device, which we call the CardioTag, that would go together with third-party devices for blood pressure measurement that are also required for GDMT, and that that would then integrate with medication protocols and lab results. And all of this would be used both in inpatient settings and then ultimately also for remote management of patients with heart failure. And strong HF was brought up earlier, and exactly as per that study, where this sort of approach led to very good results. The idea is that we could then use this device to provide the same information that in that case was provided by anti-pro BMP, but done so with a wearable device for the first time. This is our roadmap. Again, sort of a short pitch, so giving you an idea of what we're doing. Probably too focused on the technology, because that's what gets me excited, and I think that's where we're really doing something that's very innovative, and it's important in this space, I think, to provide the right information for these patients. And PCWP is really the right information. So this is kind of our roadmap. We've already done the seismic HF1, which was our model generalization study. That's what I talked about with 1,000 patients. We're going to do seismic HF2, which is going to be our pivotal. That's going to lead to our de novo submission to the FDA. At the same time, we have a couple of evidence generation studies that we're going to be running in parallel with seismic HF2. And then, really, what we have is a platform for a wearable device producing high-quality physiological signals from the chest that can then be coupled with deep learning custom algorithms that our group develops that can do waveform-to-waveform reconstruction. In this case, we're using this for PCWP, but we have ideas for other variables in the future as well, based on a lot of the literature that we produced at Georgia Tech, Northwestern, and other sites on the academic side before this even became a company. We have an awesome team. Our CEO is Amit Gupta, has a great background in software and biomedical engineering. CTO of the company is one of my first PhD students, Andrew Carrick. Mazi Edemadi is our co-founder, who's been my friend for almost 20 years now, and we worked together at Stanford. He went to UCSF for MD-PhD, and he's really a leader in AI for health. We have at least equally or maybe better clinical advisors. Liviu Klein has been a close collaborator of mine for 15 years now, and just a fantastic group. So, thank you. Hopefully, this timing was not including questions. Okay, awesome. You're perfectly on time. Okay, now we will take questions. I have the first one. When you are making your algorithm and you're using BNP cutoffs, are you adjusting for obesity because the cutoffs are going to be different for beta natriuretic peptide if you're inputting those sort of clinical variables? I want to be really clear. So, for our algorithm for wedge pressure estimation, we only use our physiological signals on our device and a gender token. That's it. We don't use any BMI, no lab values or anything. It's a purely physiological estimation approach. How easily were you able to integrate that into the EMR in these trials that have already been done? In the trials we've done, we did not have to integrate with EMR because we collect all of the data on both sides. We have our own sort of database set up for that. We have actually, with Northwestern Medicine, we've actually demonstrated the capability to integrate with EMR, and that's something that our team is very familiar with. With our existing studies, they've been purely for model development purposes and FDA clearance purposes. So, we haven't had to do that. Try to connect the dots between your clinical data and what you're trying to show with your AI, and then ultimately, the ability to commercialize. What are you doing for, what's the correlation between your wedge pressure and symptoms? Because ultimately it's the symptoms that drive patients back into the hospital, or back to the hospital for readmissions. For 30 day readmissions, they're coming in because not their wedge pressure is high, it's because they're short of breath. Or they can't sleep at night because they can't lay down. So how are you kind of connecting the dots there? Yeah, so I mean there's really good literature at this point because of the implantable devices like CardioMEMS out there that shows that there's a cascade of events that ultimately lead to those symptoms. That the earliest indicator is that people can measure is an increase in filling pressures like PCWP that then leads to volume overload, that then leads to autonomic adaptations. And then ultimately then there are symptoms that lead to hospitalization. So I think the premise in the implantables literature and in the huge trials that have been done, which luckily we don't have to, we could do them, but we don't necessarily have to because of the fact that they've already done this with implantables, is that they've shown that by then acting as early as possible to reduce PCWP, in their case pulmonary artery pressure, as early as possible, you can kind of prevent those hospitalizations because you're sort of keeping the person in a more normal volume status range. One person's wedge pressure may cause different symptomology in another person, a five versus a 20. Some people may show up with the same symptoms and causes. So I'm wondering, again, at the end of the day, you do have to customize. There are buckets, right? If you're at 20 or 30, you're gonna be coming in. But if you're at five to 10, some people may come in, some people may not. So I'm just wondering at a certain point, you may have to show the data to support what you're showing with your wedge. It's actually causing a readmission. Yeah, yeah. I mean, definitely we will have to generate some of our own evidence as well. But there's really good literature showing that if your wedge is, in this case, pulmonary artery pressure is increased by three millimeters of mercury beyond the threshold, that you increase your risk of hospitalization by a large percentage. Like these are pretty well-established numbers. And I think for wedge pressure, the cutoff of 18 is really, 18 to 20 is kind of considered to be a reasonable level. And if you're above that, then hitting the GDMT hard makes a lot of sense. If you're below that, then you're worried about drying out the patient, that's when you back off, right? But I think you're absolutely right, that the exact values may have to be. Omar, I'm just gonna make a comment so we can move on. It was really nice. But the thing that you have, the entire thing hinges upon use of wedge. And what we've seen at our hospitals, and I'm not a heart failure expert, is they get cardio MEMS, they say Champion's great. Then they realize they don't need cardio MEMS because the patients have to get a blood pressure anyway. And the goal is just to jack up GDMT as fast as possible, right, which is the unmet needs we heard earlier. So they abandon wedge pressure readings. And they feel that they're getting the same improvement. And so, again, you can have a long debate here, but I would encourage you to sort of think about how you keep a dependency on the wedge or you, from the get-go out of market, have multi-sensing. Because if the wedge is abandoned or if there's other care paradigms, you're gonna have an uphill battle. And I just encourage you to think about that. How reasonable. Thanks for the comment. Thank you. And we will move on to our next pitch. Hello? Okay, so this time it is me. When I was preparing for this presentation, I needed to find some ways to make this a little bit interesting and maybe a little bit different. So the first way is I took a picture of myself from 15 years ago and put it up here, figuring the picture now would not work so well. The second is I'm gonna show a little skin. And so this may go well or clear the room, one way or another. Anyway, I'm Michael Gorman, president of commercialization for a company called C-Medics. We have developed a non-invasive alternative to what I call an alternative to the swan gans catheter or CardioMEMS invasive technologies. And we talked a lot about heart failure up here. And so we are in the heart failure space. I'll show you how we are very different than some of the other presenters that have come before me here. So I've been to multiple heart failure conferences. And one of the questions that I ask most of the doctors that I talk to is what is the primary reason for hospital readmission for folks with heart failure? And they all tell me it's fluid overload, is the number one. There's multiple reasons, but fluid overload seems to be the number one. So one of the main things that C-Medics focuses in on is fluid overload. And we'll get to that in a second. So the primary devices for measuring fluid overload include the swan gans catheter, which actually is an invasive approach. And then you can get some aspects of that through the CardioMEMS device. So just imagine that your family member, your mother or father ends up with heart failure. So swelling of the legs, they're dizzy, whatever. And they go to the hospital. And so you get that call and you're rushing to the hospital. You wanna go see them. And they get you through the hospital. They bring you up to the ICU. And here's your parent to the left side, right? This is an extremely horrifying picture to anyone that has a relative with heart failure. And so what we decided to do was to eliminate the need for this on the left-hand side. And not quite eliminate it, because in some circumstances, it's actually needed to go in and catheterize a patient and find out exactly what's going on in there. But there's a non-invasive way to do this. However, the only way that doing this non-invasively, which is safe, pain-free, non-expensive, is to make sure that we're correlating well enough to the catheter-based approach. And so current companies that have come before us with what we call bioimpedance technology to do this have only correlated in the 60s, maybe low 70 percentile. We have studied our device over 10,000 patients over the course of 10 years without going to market, primarily because we have been working on improving the algorithm for each patient that has come along. So we've been able to get our correlation to the Swann-Genz catheter to roughly about 90 to 93 percent. And that, to me, is an acceptable correlation to eliminate the need for the catheterization approach. So now here's where the skin comes in. I have been wearing our device the entire time up here. And it's a bedside device that goes on a pole that shows waveforms. But mostly what we do is we take measurements of four key items. So we're measuring cardiac output, which has an underlying measure of stroke volume within it. We also measure heart rate. We're measuring ejection fraction. And lastly, we're measuring ZO. And so if you look at the screen and you see that somebody's ZO is 18 or 17, that's an indication that there's fluid probably in an unsafe place. And then let's say that measurement goes from 18 to 15. Well, you know that fluid's on the rise. Decrease in ZO indicates fluid on the rise. And obviously the opposite. So having those four key measurements next to one another on the screen, where actually you can see it from across the room, is extremely valuable in terms of treating that patient. So if the fluid levels are rising, obviously you want to get a diuretic into that patient and see the fluid levels get back to a normal state. We're starting with the hospital systems here as our path for commercialization, primarily because we need physician advocates to use our device on their patients, get comfortable with it, make sure that it actually works as designed. And once we are comfortable that we're seeing some commercial success throughout the hospitals and hospital systems, and we're starting in the US and then gonna go global, that version two of this is the home health market. But we're being very careful to make sure that we get it right in the hospital systems before it actually reaches the home. Because once it's in the home, then it's sort of out of everyone's control. And we want to make sure that we do this right. Now, in terms of this is where the skin comes in. So I'm just gonna show you how actually easy this is to wear and perform at the hospital. So this is a strip. It starts right here at the top, goes down to the very bottom. It's actually connected right here. If I want to take it off, I just simply take it off. And now I'm not connected. Put it back on, it's that easy. So I didn't clear the room. Thank you all for staying through that. So this is our fantastic team. We have two MDs on our executive leadership team. These MDs have been really part of critical care for a long time. They've also been part of NASA and they want to take our device actually up into space. I don't have the budget for that, but whatever, we'll find a way eventually to get it up into space. Our team is led by Deb Simpson. Deb has been involved with medical devices almost in her entire life. She's been through a bunch of exits before. You can see Kevin Ferguson and Marco Grifa. Marco Grifa happens to be the chief medical officer for the city of Las Vegas. So he's seen just about anything you can think of. And it's really great to have Mark on our team because you want to talk about real life issues. All you have to do is ask him and he'll tell you a million of them. And then lastly, we're supported by a CFO who's been a public company CFO and has ex big five background, big four, whatever it is now. I'm also ex big four. So as everybody mentioned up here, heart failure is a gigantic space in terms of its total market and 500,000 deaths per year. We're not saying that we're actually gonna prevent deaths from heart failure because people will die eventually of heart failure. But what we are saying is if you manage this properly, you can prolong their life or maybe you can avoid other complications. And as an example of that, if a patient actually starts going into fluid overload, the next thing in their life could be that they end up with congestive heart failure. Now it's really hard for them to breathe. They're really struggling and then it just goes downhill from there in terms of all the other complications. So again, if you manage this right, what you do is avoid the complications. You prolong their life because you wanna keep your family member around as long as possible, maybe. But that's really what we're after here at C-Medics. And thank you all for listening. I'll take questions. Thank you very much for being on time especially and for your presentation. One quick comment, question. One is you are going to be validating this device in the hospital, which is a sort of a different population of acute care versus you're thinking about using it in a chronic space at home. So those two patient populations may be quite different. They are different. We know that. And the other comment is how are, or question is how are you comparing your cardiac output, EF, for example, to the echo data that is readily available already? We use it all the time in the hospital. Yeah, so right now we're going through a, so as I mentioned, we've already done studies on 10,000 plus patients. And we're actually doing a study right now at the University of Michigan where the patients are in the cath lab and hooked up to a SWAN. And then we're actually putting our device on the chest of a patient and measuring that correlation right there in the lab. A SWAN is not gonna give you an EF. It will give you a cardiac output. That is true. Ejection fraction is one of these things where we will have to do continued studies to make sure that we're understanding the ejection fraction portion right now. But we feel really confident in the cardiac output and also the ZO part of it. And the ZO part of it is really, you know, the key to fluid measurement. So definitely you could see the benefit of having a non-invasive way to get information. What I'm not clear about is, and I think the study that you just mentioned that you're doing at University of Michigan is gonna be really helpful, is the ability to, you know, get a sense what is the adjustments or at least changes on the pressure. Because, and if you can, information about cardiac index and cardiac output, partly because why would you put a SWAN against any, why would you float one of those, you know, and take all the risks that are associated with it, is so that you could aggressively diurese. That's probably the main things without causing the patient to, you know, have decrease in cardiac output and stuff like that. And if you don't have that data, I don't care if it's non-invasive. I mean, that's the reason that you put someone in the ICU instead of putting them into a regular bed. So that kind of data will be very helpful for anyone for us to start seeing why, you know, it's required. And I guess we should have another conversation after you get that data. Yes, that's correct. And by the way, we have collected a good portion of that data through the 10,000 patients that I referred to before. So that's the underlying data that we kind of went an extra step. We made some changes to our configuration and we already had 510K approval, but we're going back through the process because of those changes. And part of going back through that process is re-correlating at the University of Michigan to make sure that we are where we need to be. Thank you very much for that excellent presentation. We will move on now to HRT-EX. Thank you so much. We have a problem. We have a very big problem. A trillion-dollar problem. Over half of the US population has chronic disease, such as heart failure, hypertension, diabetes, chronic kidney disease, and atrial fibrillation at a cost of over $1 trillion. Over half of these patients, such as with hypertension and diabetes, have control. We know that even a small improvement in control has very significant impact in reducing cardiovascular morbidity and mortality. I'm Paul Wong. I'm co-founder of HRT-EX and professor of medicine at Stanford. And I'm here to tell you how HRT-EX will help solve this trillion-dollar problem. Using a Stanford by-design needs-driven process, we identified a key insight, and that was that the workflow is both the problem and the solution. Electronic health records are designed to emulate the paper record system, but not to optimize workflow. At the present time, whenever we have additional information, such as a data point, physiological data point, or a patient message, the provider has to go back to the medical record, has to look at the entire medical history, the medication history, and then decide what is the appropriate therapy. HRT-EX solves that. And it does so by creating a pre-selected prescription tree that I'll show you in a few minutes. Current strategies of approaching this problem by adding additional teams of chronic disease management systems only increases the inefficiency and increases the cost of the system. What we've done is we've developed a system where we use prescription trees and, in fact, allow these to guide and expedite therapy. However, many people will say, physicians, particularly, do not like to be told how to practice medicine. That is very true. We know, however, that guideline-based therapy improves outcomes. So HRT-EX has solved this. We've combined guideline-based therapy and the ability of the physician to customize that. Today, we'll be using prescription trees. Tomorrow, we'll use artificial intelligence and machine learning to combine guideline-based therapy and what the provider's practice pattern has been. So I'd like to tell you about how HRT-EX works. HRT-EX fundamentally monitors physiological data, for example, blood pressure, blood glucose, heart failure indices, or heart rhythm. This is so that the provider does not need to monitor these. They do not need to look at the deluge of data. We know that there'll be only more data as wearables and other tools come on board. There'll be thousands of data points, really worsening the overall physician burnout and the problem of inefficiency. HARD-X takes that place. It monitors these in the background. And you can see that it integrates this with an electronic health record. It can also be a standalone system and have that information directly transferred to our HARD-X system. This is an example for hypertension management, a pre-selected prescription tree. It shows how we can escalate both the dosage, but also add additional medical therapies as needed for the patient's treatment. This shows the crux of what we do, and it shows the single-click action that's needed, that's embedded within the medical record. This allows us to be part of normal workflow. That is, the physician does not need to go back to the medical record, does not need to go back and look at the medication history. This is all provided at this interface in the workflow when, in fact, only a recommendation is needed for change. And all this information, including the reasoning for why this change is going to be made. We direct this currently to physicians, but this could be used for other providers, pharmacists, and nurse-driven protocols in the future. And yes, this works. These are the results of our pilot study showing we demonstrated a 14-millimeter reduction in systolic blood pressure, and we're engaging a 400-patient randomized clinical trial. We've also developed in our milestones an MVP, which we've shown you, and we're in the process of raising seed and Series A funding. We've received $1.3 million in grants from Stanford, Coulter Foundation, American Heart Association. We're one of nine of 160 projects selected for the Stanford Catalyst Accelerator Program with $700,000 in additional funding. We think we have significant advantage over our competition. We have the ability to have provider decision support, customizable treatment strategies, and support by the American Heart Association. Our approach is unique. We combine a guideline-based therapy with customizability, and we have a workflow focused on efficiency. We believe the future of healthcare is dependent on the ability to improve outcomes, improve efficiency, and reduce costs. This is the trillion-dollar problem. HeartX is here to solve it. Thank you very much. How are you gonna monetize this? Yes, thank you. We've conceived that we'll be starting as software as a service, as well as providing chronic disease management, and we think that really allows us to serve both and partner with different companies that are involved in chronic disease management. We're actually an excellent partner for our competitors that are really looking to provide additional services. They're the typical chronic disease management that's gone after self-insured companies, and managed care, advantage programs, et cetera. So we provide that efficiency because we think that those models are not sustainable. They use really labor, which is really not scalable, and so we provide both sustainability and scalability overall. So we think those will be some of our key partners in the initial phase. So more of a software commercialization versus a tech-enabled services type of situation. Yeah, we think that the problem is always gonna be, you know, or the question is, do you buy or build, right? For any healthcare system, whether you're one of these companies that provide these services, or you're a large healthcare system, you need a solution. We think this is, in fact, this is someone, everybody is going to face this problem of efficiency. That's really necessary for scalability, but all the healthcare systems are gonna say, are we gonna build or buy? If you're large enough, some will, in fact, build their own. We don't doubt that because we believe this is the ultimate solution that everyone will be looking for. But I think a large number of entities will turn for the buy solution, which we'll provide to them. Thank you very much for that. And we will move on to our last presentation from RCE. Hi, everyone. Thank you for having me here. And great presentations on heart failure and also being at a heart rhythm event, I'd like to say we're a company that looks at coronary artery disease. And this is a precursor to many damages that happen in the heart, including arrhythmias as well as heart failure. I lost my dad to a heart attack seven years ago, and it was a silent one in nature, and it left me devastated. And I kept wondering how could someone have not been able to pick this patient up? My dad was a second heart attack patient, so he already had a heart attack, established care, had disease, and then he had a silent heart attack. He didn't feel chest pain, so he couldn't go to the ED. And I kept wondering, was there gonna be another solution until we conceived remote cardiac enablement? So today, heart attack is diagnosed in the hospital. And the only way you can diagnose this is when patients show up and you do different tests like EKGs and you draw blood and you do serial draws and you look at troponin changes and that's how you diagnose heart attacks. What we came up was a technology that could scale from the hospital into the home where patients can be monitored from their baseline. But also patients, when they show up to the emergency department, can be diagnosed for a heart attack very, very early. So a little bit more into that, but prior to that, I'd like to show you some of the early work we did looking at the current standard of care, which looks at cardiac troponin, which is the biomarker for diagnosing myocardial infarction. And the way we did this was using an optical-based technique. It's an infrared spectroscopy modality, which is quite versatile in the chemistry space. It's used in laboratory medicine. And the modality we used was, if we were able to have a photonic exchange of light between the, through the skin with molecules of interest, such as cardiac troponin, and some of the other biomarkers mentioned here, like fatty acid binding protein, pentraxin, families of BNP, et cetera, we will have a better sense of what is happening physiologically with the heart. And from our wet lab studies, we were able to identify characteristic fingerprint regions in the optical spectrum that could identify these proteins. And from there, we came up with a non-invasive variable, which you see here. And the way this works is once you are ready to measure this on a patient, you bring the patient's wrist onto the sensor surface, secure the patient's wrist with a wristband, and leave it on for three minutes. In three minutes, we're able to extract an infrared spectral scan through the skin. It's a continuous scan. And during this time, we're able to take this data, send it to the cloud, and come back with an instant diagnostic report. And what that means is in three minutes, you have an early assessment of what is happening with the heart. Our early correlation studies were obviously very focused on high sensitivity cardiac troponin I. These were published in journals and places like Nature Communications and Circulation and presented at AHA. Again, these are comparisons with just high sensitivity cardiac troponin. An early feasibility study was done in 238 patients across five hospitals where a machine learning model was able to demonstrate a very strong discrimination between elevated cardiac troponin I versus non-elevated cardiac troponin I in patients suspected of myocardial infarction. We also showed good phenotypic association with severity of stenosis in the coronary vessels as well as wall motion abnormalities. And from here, we started an investigation into understanding our signal. We put our signal on a continuous basis to really understand myocardial infarction or even better acute coronary syndrome. And what you see here on the left is a graph that shows a serial recording. It's a raw signal recording. When you place this sensor on a porcine subject and you create a myocardial infarction. So what you see on the left of the black first arrow is a baseline collected by our sensor. And then you have a 90 minute angioplasty. So 100% block in the LAD of the coronary arteries. Now this is a catastrophic model. So right after 90 minutes, you deflate it and you watch what happens. What we see is a baseline rises right off very shortly, very, very shortly after balloon inflation. And right after deflation, you see a market jump and then it steadily continues to rise. And when we took serial blood samples throughout this procedure, before inflation through the procedure and then post deflation, we were able to see a rise in cardiac troponin I post deflation. And this is very important. This is what literature shows us. And this makes sense. Once you revascularize with the flush, you get all the cardiac troponin down the circulation. But what happens between inflation and deflation? Why do we see an inherent rise? And we were able to do additional biochemical validation to look at those early markers that are released in small vesicles and that we were able to pick up. And this was corroborated by further studies in the cardiac cath lab, where we looked at different patients who are coming in for diagnostic cath procedures, emergent MIs. And we were able to see a very sensitive and highly specific signal, which corroborated with the coronary insult. Some of these cases were already presented at the ACC and DCT conference. The one on the top is an aspirated thrombus that was dislodged as part of revascularizing a total occlusion. Missed on initial angiography and subsequently picked up on angiography. You can see the signal rise, that market peak, and then come down. The one on the bottom was a catastrophic case. This patient could not be revascularized. The signal stayed up and the patient died two days later. Interestingly, these are some other studies. This is out-of-hospital cardiac arrest patients. Patients came in, they got our sensor, three minutes at time zero. At that time, they took a blood draw, and then an hour or two hours later, they took a second blood draw. Our single sensor reading at three minutes at time zero at the door matched the delta change. So our prognostic value of that is significant towards myocardial infarction. Recent study in 840 patients across 10 US hospitals. This has been accepted for publication, not published yet, but basically a very strong discrimination between elevated cardiac troponin versus non-elevated cardiac troponin. This is a recently completed study. Again, you see the time-adjusted difference, so the delta changes between blood-based sensor and our sensor, very good correlation. And what's very interesting in this study was what happens if you were to leave the sensor on beyond three minutes? And in the first three minutes, you already get a very good diagnostic accuracy for acute coronary syndrome. And what this means is this is a single measurement of our sensor, no modalities of gender or EKG or anything else, but a single readout from our sensor by itself. When compared with the diagnosis of acute coronary syndrome based on clinical evidence of EKG symptoms, biomarkers, angiography as needed, was very good. And when you left it on for 20 minutes, you even got better. So it's very interesting how this could start to pan out really in the world. Something that we've been recently working on towards is how do we bring this all together? When we look at the spectrum of, in this case, about 1,300 patients where there is obviously a lot more non-ACS versus ACS, this is what happens in all comers. And if we were to set the threshold of our algorithms to about a 99% sensitivity, then we're able to get a three-minute disposition through negatives, through positives for about 60% of these patients. So this might have some relevance in a busy emergency department, crowded emergency department, where most of the patients, 60%, are going to be non-cardiac. They can be acid reflux, chondritis, anxiety, upper back pain. And that leads us to envision or start thinking about potential optimizations in the ED. And this can allow for potentially early rule-outs of those high volume of low acuity and low reimbursement patients, which is of value to the hospital, while fast-tracking some of these early access or early diagnosis of ACS, which qualifies them into a higher reimbursement bracket. You can push these patients through, and this really brings in more better capacity, better flow, at the same time providing good cost savings and revenue. Our business model in a huge market with 10 million chest pain patients running about 25 to 30 million tests per year will be set at a $50 per test. This has been validated with about 30 hospitals. That would be a good price point to get at and would be a volume-based year pricing model. Where we are is looking at a pre-sub and considering potentially a pathway towards early assessment for acute coronary syndrome in acute care settings, starting with the emergency department. And we're looking at a pre-sub to review our pivotal endpoints and thereafter go towards market. This would not be possible with a very strong team. I've had the opportunity to work with really good doctors like Dr. Sengupta, Dr. Chaffee, Dr. Stone, and a lot of experts here in the room as well. And this would not be possible without the support and truly innovation at heart here. Thank you. Thank you very much. So if you have to, well, actually, maybe you've done it already. What does your FDA label look like? Yeah, this is a very interesting question. I don't know if I can go back here. And we have had a chance to think about if we are able to track biomarkers in very early, what is the most value to the doctor and the patient? And doctors want to understand the underlying state of the patient and do something, you know, if they know about the disease. And we believe something like myocardial injury or acute coronary syndrome, just an assessment of that early on, can be of value to the clinician. So if you know this patient is not having some cardiac burden, then you can manage, the ED doctor can manage them differently, versus if it is definitely something to do with the coronaries, pull them in, do your imaging, and provide care, whatever that might be. Just so, is your clinical studies now really focusing on like certain patients that, you know, in the ER, you need to rule out? For example, you have a patient in their 20s coming with atypical chest pain. I mean, that's the first thing that you want to rule out. Are you incorporating a lot of these type of atypical rule out patients in your clinical trials? Yeah, so, I think that's a good question. Yeah, so, our studies have to be all comers, that way we don't miss, and you'll see in this study, design 1500, it's a larger study. The prior study is 840, and the 500 patients were also very heavy on the non-ACS patients. The 840 had a lot of healthy people. We sourced the blood samples, we sent it to University of Maryland as a core lab, we insured troponin, antiprobiotic creatinine, HbA1c, so that we knew what was truly negative. So, it's very important. Can this be used as a rule out? I think that has to be seen by the ED doctor, but what we can do is, from a clinical performance, give a sense of true assessment of the heart. Excellent presentation. One question I have is, so, ACS can be obstructive coronary disease, but it can also be non-obstructive coronary disease, and we know that that population is certainly rising, and in women, two-thirds are gonna be non-obstructive. And so, my question is, are you dividing up some of these patients that you're seeing and correlating with their cath findings or CTA findings to see if you can differentiate? That's a really good question, and there's so many ways to answer that. I think in the 840 patients, we had almost like 52% female and male, so we're doing a lot of moderator analysis to see. Good thing, and as you mentioned in one of the other presentations, antipro-BNP and differences in gender. Even troponin S1 marker has it, but the change is very characteristic. If you see an acute rise or fall, that has clinical relevance. Now, is there a difference between type one, acute plaque rupture, versus type two, non-obstructive? That has to be seen, and that may be a very important thing that we can provide in that 20-minute acquisition. So, I think those are things. Now, the other thing is in the 500 patient study that we just wrapped up, we have about 450 coronary angiography data, 300 echocardiogram, and all of them have 30-day follow-up. So, there is some level of deeper analysis, secondary endpoints that we're trying to understand. Maybe we might be able to find more in these sub-cohorts of populations, but again, we might have to step into that slowly. And women's heart disease, as I understand, will be a very complicated place to solve, but at least if we can just understand injury and bring them in, and then let the imaging be interpreted internally or biomarker analysis or whatnot be understood by the clinicians better, that would be a good start for now. Thank you very much. This has been an outstanding presentation by all of you, and I wish you all the very best of luck. And like I said before, you all are winners. You're on the stage, you're presenting to us, and it has been a remarkably fast hour and plus hearing these innovations from all of you. Best of luck. Thank you.

health tech startups

heart care

artificial intelligence

virtual monitoring systems

AI-powered virtual clinic

medication therapy management

non-invasive device

remote monitoring

wearable sensors

chronic disease management

cardiac biomarkers