Mohamed Sadeq Ali, AccurKardia - Spotlight Studio | LSI Europe '23

Transcription

Nick Talamantes  0:17  
I'm here at LSI Europe in Barcelona with Sadeq Ali of AccuraKardia. Sadeq, thank you so much for joining me again in the studio.

Mohamed Sadeq Ali  0:23  
Oh, thank you for having us.

Nick Talamantes  0:25  
Yeah, I'm looking forward to hearing what's new. So why don't we just jump into that? You know, we we met last year in LSI Europe in Hertfordshire. You told me about what your company is doing? What's changed in 12 months?

Mohamed Sadeq Ali  0:38  
That's a loaded question. Actually. We spoke, I think in September last year, right. So back then, maybe even a couple of weeks after we spoke, we did our first for FDA filing for, for our core product AccuraECG. And we received a response letter from them in, I think, right before Thanksgiving. And I think the response letter was the first indication that things were going well, because it wasn't an 18 page response letter. It was a, it was a letter with, you know, some very good questions, but we knew it was manageable in terms of what they were asking us to provide them. And so that, in some sense, put us a good move into a really good mode going into the beginning of 2023. We had JP Morgan, which was a great week met some some really great strategics and investors there. And then the following week, we heard from the from Mayo Clinic that we were going to be invited to join Mayo Clinic platform accelerate. So the your, the end of last year and the beginning of the year just started off with with a lot of momentum,

Nick Talamantes  1:56  
I just want to comment on the momentum. And you guys have accomplished a lot from what I've been following in the 12 months since we've seen each other I know that you were also at LSI event in the USA this year. How was

Mohamed Sadeq Ali  2:09  
That was a lucky charm event for us, by the way, because while at the event, I think the second day of the event, we heard from an our first institutional that they were ready to give us a term sheet. So the you know, it was unrelated in some sense. But you know, there was definitely the karma coming from the event that that fell into it. But yeah, that event was was was was where we basically got the momentum as far as fundraising was concerned.

Nick Talamantes  2:44  
So you're speaking to the FDA last November. And it's my understanding that this June, you guys receive regulatory clearance. What does that mean now for you guys?

Mohamed Sadeq Ali  2:55  
Well, it means what I can talk about the product. Let's do so. Yeah, I think the the clearance we received was for a basically a class III medical device. It is a heavily regulated device because it can indicate certain critical arrhythmias. So basically, we got the approval for 13 different arrhythmias, including some, you know, important atrial arrhythmias such as afib, and a flutter for stick vav block as well as been trickly. arrhythmias like ventricular tachycardia. Given the type of arrhythmias we we are diagnosing, they're very serious arrhythmias, it was the bar to proving that our analytics work was set pretty high. And that's really what took us from, I'd say, November through to the end of May, we needed to just provide the FDA with a lot more data than we'd originally provided them with. From our perspective, B, it provided them more than what our understanding was predicates had provided them with. But I think the FDA is bar has gone up a bit as well. And well from we needed to comply with that. So it took us about a good five to six months to do to provide them with all of the information that they were looking for. Which involves doing a study with a panel of cardiologists to make sure that they first agreed on what arrhythmia existed in, say, a 10 second ECG strip, which is not easy in terms of having three cardiologists agree on on on something like that. And then comparing that against what our analytics engine was diagnosing. So we we took the stand that you know, whatever the cardiologists annotated, was the right annotation if all three agreed and the system was compared against that. So that process really is what extended it through the end of May, we sent the response in I think, right before Memorial Day. And like literally a month on, we got the approval.

Nick Talamantes  5:10  
So for your peers in the AI analysis and big data, I actually think you don't like using that word. Right. I believe we talked about that last time.

Mohamed Sadeq Ali  5:19  
Yeah, I mean, I think the there is definitely not not to totally underplay the value of big data, it's, I think what matters more in, in in our space is honestly the quality of the data. And if possible, the ability to get longitudinal data, which is often hard to get, which is where the the Mayo program was very useful. Because we, of course, by then we didn't need the data for our core product, because we'd already had all the data we needed, we had done the submission to the to the FDA. But it turned out to be very useful from the point of view of building certain new algorithms, which are, I'd say, relatively not novel in what they can do. We, from our perspective, we have sort of two core value propositions. One is in the space of the arithmetic section, which is traditionally what's done through ambulatory Holter, monitors, MCT, devices and patches. And that's foundationally, the core of our business, that's where we are focused commercially at this point. But alongside that, we've developed this thesis that the ECG signal is a potential biomarker and a very broad biomarker at that. And the there are, there are multiple reasons for it, I can touch on those, but essentially, the, what's what's bearing out now is that the ECG signal can be leveraged for things well beyond what it's traditionally been used for. And there's a few reasons why why that's that's possible today, AI being being among those. But, you know, we we then started to use the data from Mayo for diagnosis of things like hyperkalemia, for example, which is an disorder that matters a lot in nephrology, right, so if you have hyperkalemia, the potassium levels in your blood go up. Traditionally, the way the testing has been done for that has been through a blood test, what we have done is we've been able to indicate a, a, a moderate, or high level of hyperkalemia, using the ECG signal, for something like that you do need a significant amount of data. But I wouldn't call it big, big data. Still, it's not in the same range of data that you hear off, for example, with large language models, we're talking about data from the United entire internet, or it's not that scale of data, it's still, you know, relatively small amounts of data compared to that. But high quality data with like, for example, it's not good enough to have the ECG signal data, you also need the blood test data to compare against, right. So those are the type of things that that matter. So I'd say I'm still definitely much more of a proponent of the quality of the data than the volume of the data, particularly in healthcare.

Nick Talamantes  8:30  
You've given me a lot to unpack there with all that. And I want to touch on the a few of the things you've just mentioned. But I do want to come back to something that stood out to me. You mentioned that your transparency with the FDA about how your AI algorithm works, it sounds like it was instrumental to your success. I

Mohamed Sadeq Ali  8:50  
just have a complete candor here, we actually, for our first approval, because our algorithms were performing very well, even without, as I mentioned, last year, we were using AI in a very surgical way. We are in in some of our products, but in in the product that has received FDA approval. Because the performance was already relatively high. We took a different path, we decided to focus on the analytics piece, with full transparency without the AI elements in it primarily because the performance was high enough to support that our sensitivities for arithmetic detection are over 96% specificity is over 99%. At those numbers, adding the AI bid was only going to complicate the approval process. So our our strategy was extract the AI bit out, get the approval and then build upon it, bring the AI bit back in and get additional approvals for certain areas where we do need the air and the AI bits. So the approval that we received in In June, was for the core analytics piece without the AI bitched for complete candor.

Nick Talamantes  10:07  
So AccuraECG is cleared. What device? Can we use it on today? How does the patient get access to it? What does that process look like? 

Mohamed Sadeq Ali  10:16  
Yes, today, the way that would work is a hardware manufacturer, let's say an ambulatory device manufacturer would have to integrate our analytics capabilities into their offering. Because we are not going directly to a cardiologist or to a patient or anything like that, we're essentially providing a Holter manufacturer or a or an MCT device manufacturer, access to our analytics capabilities, so that they can enhance their their reporting. I'm not sure if you know how that workflow works today. But essentially what happens is, if you say go to a cardiologist, they give you a holter to wear, you wear that hold till you may wear that through for a week, a couple of weeks, you go back, you your hand in the Holter, the data from the Holter gets uploaded to the cloud gets analyzed initially by their software. But then there are these cardiac technicians, both in the US and offshore, that review the reports, annotate the reports. And and finally produce a result that whole process can take from the time the cardiologist gets the data back or the Holter back from the patient through to the time the patient is being provided a report can be anywhere from two days to a couple of weeks. And that is a very, very long process. And it is an expensive process. Because there's a lot of cardiac technicians involved. And there's a lot of sort of human labor involved in that process, what we're effectively doing is we can reduce the time spent by a cardiac technician on each of those reports by about by up to 70%. So a single cardiac technician can go through, you know, even for longer duration reports, you know, many more reports, and other than he or she could without our software. But our software isn't meant to go directly to a cardiologist in any way is meant to to support the whole two companies, the patch companies, if it is truly to MCT's as well as event event monitoring companies to be able to integrate our offering into their solutions.

Nick Talamantes  12:35  
Is it going to be compatible with smartwatches and things of that nature? Is that a discussion you guys are having? 

Mohamed Sadeq Ali  12:40  
Yes. So that that's probably going to take us lightly. So we do have a prototype of that built already. Effectively, it's the same analytics engine in the backend, it's just, you know, the the device, ingesting the data, is what's called a has what's called a dry, dry lead, whereas our approval at this point is for a wet lead. So if we are going to support smartwatches, the first step towards that is to get approval for for dry leads. That's also critical for a lot of, of you know, that's that's also critical. Outside of smartwatches. There are other use cases in which there are dry leads as well. But yeah, I mean, our goal is to get approval for dry leads to be to be available for for not just the Apple Watch, but a lot of clinical grade smartwatches that are that are coming up as well.

Nick Talamantes  13:31  
Yeah, you would have access to a massive patient population. At that point, I believe there's some 50% of people wearing a smartwatch these days.

Mohamed Sadeq Ali  13:39  
Yeah. And many of them are now being embedded with ECG sensors. So ECGs are literally I mean, this is the reason why we believe that, you know, ECG has such a large value as a biomarker. And when we see a biomarker, we think of a biomarker, what we call a planetary scale, right? If you think of planetary scale biomarkers that have broad diagnostic coverage, that is that can detect multiple conditions, really three come to mind and to come to mind to the general population that the third that we are trying to in some sense of Angela's accg. The first two are blood and DNA. But there are some significant disadvantages with blood and DNA. Neither of them can be remotely or continuously monitored. They can be pretty expensive, even you know, blood tests can be expensive, but gene sequencing and DNA tests can be a lot more expensive. The time taken to get the results is a lot longer. And as importantly, it is really difficult for a non expert to administer, let's say a blood test, right? You need to go in and get that done. Whereas with an ECG, you don't need any specialized training to record an ECG on your Apple Watch, right? So if the ECG can become that sort of third big biomarker for broader diagnostics coverage, The scale that ECG can achieve is truly planetary, because as you mentioned, right 50% of the US population probably has some kind of a smart device smartwatch, it's only a matter of time before the rest of the world has that as well. And then you need to have a solution with the features that allow you to get that scale and get value from that scale. And there are certain key elements that a solution like that needs to have. And we believe that even our very first product has all of those key characteristics. So we're very bullish on this idea of ECGs, or biomass biomarker, and being the company that's able to deliver that.

Nick Talamantes  15:41  
So ECG is vital to identifying patients who have arrhythmia. But it's also I imagine there's a lot of value that can be added with, you know, following these patients as they're being treated to and understanding their response to treatment, is that also what your platform is capable of?

Mohamed Sadeq Ali  16:00  
So it depends on the integration, right? It depends on what in the in with halters. With patches with when monitors, it's unlikely that you're going to wear it long term, right? You're gonna wear it for maybe a few days, a couple of weeks, a month at most, right? In order to do continuous monitoring, which is where the value of I think something like ECG truly comes into play, you do need to have a device that's more of a wearable, right. But once you have that wearable, you can not just monitor the individual from the point of view of arrhythmias, but you can go well beyond the arrhythmias. I'll give you an example of an area where we're truly excited and are spending a lot of our time. Right. So at the Mayo Clinic, we developed as part of platform accelerate, we developed a an algorithm for hyperkalemia detection. So what is hyperkalemia? It's high potassium levels in the blood. Now, you might think of that as being okay, is that somebody? So Tarik, let me let me just give you some numbers, right 750 billion a year is spent by Medicare in the US every year right off that 25% spent on chronic kidney disease, chronic kidney disease, you know, a large percentage of people with chronic kidney disease, you also end up with end stage renal disease, about 40% of people with end stage renal disease die not because they cannot get a kidney replacement, but because of hyperkalemia. So just to give you some context of scale, the scale for hypokalemia detection, if it can be done through an ECG is massive, right. In terms of the current gold standard, it's through a blood test. But the issue with a blood test is hype. Well, there's two issues. One is it takes longer to get the results. And it's more complicated, right, you have to go and have your blood drawn can take easily four hours before you know what's going on. The other is hypokalemia is potentially fatal, you know, particularly high levels of hyperkalemia, it needs to be acted on immediately, you don't have time for a blood test, right? A lot of times that you just based on symptoms, you're sent, you're not even sent to a dialysis center, if you if you show up at a dialysis center saying hey, I think I have high potassium, you are immediately whisked off to an ER and an ER visit costs a lot of money in. So from that point of view, the benefit to the entire health system in some ways, if hyperkalemia particularly when it isn't dangerous, yet, right can be can be identified makes, you know be hugely beneficial. And if you can monitor that through a smartwatch, nothing like it. So even with the algorithms that we developed, we've done it using lead one data only. So why is that important? Well, the Apple Watch works off of the lead one, it doesn't it's not a full 12 lead rights, it works off lead one. And our algorithm at least we tested we did a retrospective study with about 20,500 patients with Mayo Clinic data as part of the program. And it performed brilliantly we had you know, sensitivity is of about 81% specificities around 83 Which specificity basically limits your false net false positives. And and an AUC, which it's something that's used to calculate how efficient you are not just how efficient and how effective your algorithm is overall, balancing both accuracy and false positives. Or I should say, the true positives and false positives. And that number comes out to around 89%. And we've we've already done some validation outside of Mayo Clinic data and the numbers are largely holding. So we're as excited about some of these new things that we're developing as VR about the arrhythmia use case and to your point, you know, once once we can start integrating these algorithms, not just with the more traditional devices, but with smartwatches. And, and, you know, other types of wearables. A whole universe of opportunities open up. 

Nick Talamantes  20:25  
Yeah, absolutely. You know, we're all familiar, I think, with continuous glucose monitoring solutions. We're talking here about continuous electrocardiogram monitoring, yeah, and product category?

Mohamed Sadeq Ali  20:35  
And frankly, speaking, right, like glucose monitor is going to give you one piece of data that is associated with diabetes is obviously a very, very large problem, but even diabetes, just to give you some, not to, not to overwhelm everyone with numbers, but if you think about the numbers of diabetes, right, there's about I think, close to 6% of the world population that has diabetes, of which a very large percentage is type two diabetes, the US alone, I think it's something like 32 point 3 million, which is about 10%, or one in 10 people in the US, right? Type Two Diabetes leads to chronic kidney disease. Chronic kidney disease leads to cardiovascular disease, people with stage four and five, chronic kidney disease. 40 to 50% of them die due to cardiovascular complications, everything from fatal arrhythmias, through to through to heart failure. So frankly, speaking, you know, these metabolic diseases and cardiovascular disease are heavily interrelated. So when you're monitoring someone for cardiovascular disease, or you're monitoring someone for hyperkalemia, you're monitoring a spectrum that goes across different conditions, these comorbidities play off each other. The impact goes well beyond purely cardiovascular, or purely metabolic diseases. 

Nick Talamantes  22:05  
It is truly exciting to hear the progress you guys are making.

Mohamed Sadeq Ali  22:09  
Yeah, and those are with ECG. The reason we're so sort of enamored by the ECG signal is because it literally is, in some ways, a treasure trove of data. And it's also very unique to an individual. I don't know if I mentioned this last time we spoke, but every individual's ECG is different to a point where ECG can be leveraged as a biometric there are papers written about this, people have built prototypes on on this, right. So the changes to an ECG, then become critical as both an indicator of health and disease. I'll give you an example. A few years ago, I ended up with a Delta variant of COVID. And I was tracking my ECG meticulously throughout, I could see what happened when I was starting to fall sick to when I was actually sick to when I started feeling healthy again, right? So you can those changes show up in your ECG. And if those changes are visible to me, as an untrained user, in some sense, with my naked eye, can you imagine what it is to a trained algorithm that can look at, you know, 100 different parameters all at once. So the ability for the ECG to be able to even predict what might might be about to happen is enormous. We, you know, we are, it's still preliminary, but we have seen in certain disease conditions that have actually fatal disease conditions, there are no cures for some of them, but there are drugs that are starting to come up to extend life, we are finding ECGs picking up those markers. This is why we call it a biomarker years before the condition actually manifests and the person ends up in needing kit, right, by which time it's even more difficult to do to give them that care. That's why we are very, very excited about this. Right. It's the closest comparison to what ECG can do. Right? And the the electrical signal from your heart the closest comparison, and of course, I think, you know, there are limits to that comparison. Is DNA, right? Like the DNA can tell you a lot about how an individual's health might progress over time. But the DNA DNA is interesting, because with DNA, you know, it's not just your your, what, whether you have a gene or not, that matters that your epigenetics matter, as well. So the DNA that is in some sense, the hardware and the epigenetics is the software that triggers the manifestation of a certain condition. So you may have a gene for something but it doesn't mean you're going to get that right with ECG. It is showing that you're actually in some sense that condition is starting to manifest right. So yes, it may not be able to tell you that at birth, which would, you know, the DNA can but as that risky condition starts to approach, you start to get notified, right you start if if we can harness that data, you know, everything we've been speaking about for years in terms of Preventative Medicine, predictive medicine, all of that becomes possible, right? So effectively, you're then able to truly transform the way in which healthcare is done. That's ultimately the vision here, right the ability to to be able to take healthcare, which is right now a lot more reactive and based on symptom solution to Alright, symptom is starting to manifest, it may still be years away from from actually becoming a problem. But let's do something about it, let's start to at least monitor it more closely. If you even if you don't do anything about it, and doing it in a way that doesn't raise too many false flags, and you know, it becomes a situation of crying wolf. That's why in something like this, as important as it is to optimize for, you know, true positives and true negatives, it is equally as important to optimize for false positives. So people are not constantly alarmed, and doctors are not constantly, you don't want an already burdened health system to be even more burden, you want to help the health system react better.

Nick Talamantes  26:18  
You guys have developed a solution that can closely follow that epigenetic expression of diseases. But based on a technology that, you know, again, it's a few percent of people have, 

Mohamed Sadeq Ali  26:31  
It's also been it's not a new technology, right. ECGs have been around for over a century. It's just that the capabilities that were needed from the point of view of specifically the AI capabilities, the compute needed to run the AI all of that was just not there until now. Right? So now that it is here, I think we can do a lot more than then, at least at this point we've, we've thought that we could with with ECG, it's, it's definitely going to remain a critical part of the traditional cardiologists toolkit. But I think it's going to start to do things well beyond what what it was originally meant for, you know,

Nick Talamantes  27:13  
maybe if we shift gears a little bit here, this is your third LSI. If I us understanding correctly, why do you guys keep coming back?

Mohamed Sadeq Ali  27:24  
Honestly, we've, we've, we've gotten a lot of value out of it in terms of like, I'll give you an example from the last LSI, right, the idea to so the hyperkalaemia model we already had in mind, we had, we definitely wanted to pursue that path. But we met a, an individual, I'm not going to name the company at this point. But it was a notch tower manufacturer, I guess, that basically said, Hey, if you can use ECG data to tell us if you know, someone might be suffering from ABS, which is aortic valve stenosis, that could be very, very useful. If you think about it, you know, again, like, you know, it comes down to the ease of getting an ECG was how a valve issue is today. Determined, right? So if someone has ABS today, the gold standard for that is an echocardiogram. But an echocardiogram is administered like once every four years or something like that ECG committee administered, like, literally in a regular doctor's visit, right. So from that point of view, if we can detect ABS through an ECG, the value of that both from the point of view of the patient, as well as from the point of view of, you know, the the device manufacturers is, is very high, right? That didn't come to us from arm, you know, sitting around talking on it in a roundtable of brainstorming ideas that, came being at LSI, a Dana Point and speaking to, to a tablet manufacturer, right, so the value that we get both from the point of view of talking to, to, to strategics, from the point of view of knowing where the med tech community stands, in terms of its its latest and greatest thinking in and in terms of just meeting a lot of, you know, great and friendly people has just been there. And when I say friendly, I'm not talking about just Hey, hi, how are you, but also the type of people that can make the right introductions and all of that has been has been very valuable.

Nick Talamantes  29:27  
So a lot has happened with Accurkardia in the last 12 months. What's the future look like for you guys right now?

Mohamed Sadeq Ali  29:35  
Yeah, I think. So. We honestly it's a good problem to have. But essentially what happened was we closed our seed round, literally three days before we got our FDA approval and our expectation and what we'd communicated to, you know, one of our major investors in the law Last round is, is the is a venture fund that was built by by Banco popular. And we were, you know, basically talking to their board right before the closing and mentioned that, hey, it could be six months to nine months away, right? That's really that that was our expectation. Yeah, not not three days after closing, right. So frankly speaking, we had raised enough from the point of view of, of building a little bit of a war chest, but also continuing r&d And all of those type of things. We weren't necessarily looking to aggressively commercialize, immediately, that was not the goal of that raise. And while there was a lot of exultation that we got the approval in them, my immediate reaction and one's immediate reaction ones the CEO was, like, it was almost like a oh, shit kind of moment. But now we need to do truly like, you know, pull up our socks and start to commercialize. And the first part of that process is to go out and raise more capital so that we can commercialize more aggressively. So to that end, we are starting a Series A round in, in the next week or so, while I'm here in Barcelona, one's doing a road trip, just meeting certain people. And so just another big thing that happened towards the end of August, is we were selected as quarterfinalist for the digital health awards that are going to be announced at to help in Vegas, the finalists have not yet been announced, the winner will be announced had held, we're hoping we at least make it to the finals. But essentially, one was in, in San Francisco and on on Tuesday and Wednesday, are actually Monday through through through Wednesday, New York today and tomorrow. And then Seattle over the weekend, right. So he's like literally criss crossing the US before heading back to to Puerto Rico for a bit. And I'm obviously here, so we're getting ready for what's going to be a very long fall. And our immediate next step is to raise the Series A round in in the fall, while continuing the commercialization efforts. So we aren't waiting to raise the series A to do the commercialization. But we are very realistic about what we can achieve in the very near term. So we are basically there are some some interesting partnerships in the play with with our core analytics engine, AccuraECG. And, you know, I'm going to be spending a lot of effort on that in the fall. But alongside that, you know, financing is going to be a key element of what we spend our time on. Beyond that we've developed, as I mentioned earlier, we're very, very bullish on some of the algorithms that we've developed, specifically the hyperkalemia detection algorithm. We've done some retrospective studies already, we're extending that out to do additional retrospective studies with, you know, large hospital health system in in Texas. And hopefully shortly after that, or maybe even in parallel, do a prospective a prospective study as well, and start looking at the regulatory pathway on that as well. So I mean, there's a lot of work, essentially this fall between clinical r&d, fundraising, and of course, commercialization of what we've already achieved with AccuraECG. 

Nick Talamantes  33:33  
It's been great catching up and hearing what's new in the last 12 months since you and I last spoke, and I look forward to the next six to 12 months and what it holds for your company. Sciatic, thank you so much for joining me in the studio. Thank

Mohamed Sadeq Ali  33:46  
you for having me. And it's always a pleasure to talk to you. And as I mentioned before last year, it was awesome. It was and this is it's been solid so up to this point. Thank you