#442: Can ultrasound + AI predict premature births?

33:35Well, who is it that said the more I practice, the luckier I get? Yeah, yeah, exactly. I mean, sometimes you just have to be smart enough to realize that you're just fundamentally on a wrong track and maybe just kind of maybe pivot. But sometimes you just, if you know, in your heart, you know there's a solution here and it can be done. Sometimes you just got to keep going. You just got to try harder than other people. Yeah, it's a reasonable thesis that on a long enough time horizon, you're going to find the correlation of these images to these outcomes. That's a reasonable hypothesis that then over a long enough time horizon, you plus AI is going to work that out.

34:14I've got so many questions on this. first thing I want to say is is is thank you for sharing about your family I know that's not that's that's not a given it's not an easy thing to do but it's it's it's it's wonderful context for the the fire that was lit within you that you did just commit to this and I think that's incredibly useful context so I appreciate I genuinely appreciate you sharing that it's also quite um yeah it's always quite real for me my mom I'm an only child and not through want of trying as my parents would say my mom had multiple multiple multiple miscarriages um and so it's something that I was always you know you know very aware of and and you know affected our family and I've got a one-year-old and so all of the birth and the pregnancy is always so fresh in my mind and all that emotion.

35:07And so I know and I understand and I can sympathize, not empathize with that fire being lit within you to solve a problem. So I genuinely appreciate you sharing that. I have so many questions on this. Oh, yeah, go for it. With the tech and the process and the medical stuff. How about I let you ask? Because usually people have a million questions at this point. So yeah, feel free to ask away because uh okay uh i think that'd be useful when you reverse engineer this what is ai looking at measuring and where where is it what data points sure is the correlation is the answer between these images and the date yeah what's the correlation there what what is the the algorithm i guess yeah a lot of doctors ask that question because you know especially since a lot of them said we've been looking at these images for decades.

36:05Well, this is the thing. This is what I'm thinking, yeah. Yeah, so yeah, that's like their immediate question now because now I've proven it does work. So now they've got to go to how does it work because saying it doesn't work doesn't work anymore, right? So basically, the AM has made a lot of discoveries. Currently, or previously, there was really only one aspect of an ultrasound image that would be useful for predicting premature birth, and that was the cervical length. And a short cervix, you know, has a decent probability of going to a premature birth. But that only applies to a small proportion of women.

36:38So that's just one anatomy, and that's all they really had that they could quantify in any way. I gave the AI all of the images that the doctors take, everything from the baby's feet to the ovaries to just, I didn't know where half of these things were. And sometimes it's anywhere from 20 to 100 images of different things that the doctors take pictures of. I didn't know what any of it was, so I gave the AI all of it. basically, and you have to sort it out and figure out what matters and then be able to predict a delivery date from this. And so I, you know, I let, after training on millions upon millions of images, the AI actually found eight anatomical areas that can predict, you know, delivery dates instead of just the one.

37:18I mean, well, actually the cervical length doesn't actually predict the delivery date even, it's just good at predicting premature births, which is a little, you know, get a little bit of a date, I guess, in that concept. But anyhow, so the, what it discovered was a lot of anatomy. The ovaries, for example, is shocking. No one would think that would be predictive, but it appears to be decent at predicting a premature birth. I'm sorry, miscarriages. And then there's the placenta, the uterus, the umbilical artery, the lower urine segment, the cervix. And then there's a couple of a baby's anatomy.

37:58It's almost entirely the mother's anatomy, except in a couple of rare circumstances. But because right now people look at the baby to get an estimated delivery date to identify how old it is. But if you want to make this prediction of when the baby's actually going to come out, it's the mother's anatomy that's actually by far more critical. But there's a couple of anatomies in the baby that are predictive, but they need a lot more research. But it appears the baby's heart, the heart rate measurement appears to be predictive of a future situation where the doctor may need to induce due to high blood pressure uh this needs further i mean a lot of this needs a lot more research to really make definitive statements about but the the as found some powerful correlations and i and i think those correlations warrant further research for sure so don't uh don't assume like this is all settled uh science this is extremely early days stuff no no sure and but but that's an interesting point isn't it that this being this being a frontier this being an area where we're learning new information and it's and it's nice obviously as a clinician myself to hear anyone uh caveat their claims by saying look this is this is an area where more evidence is needed it's always a it's always a very sort of exhale moment for any clinician to hear that as someone that's not trying to over claim and I think that that's probably my next question actually so um a complex of questions here sort of how accurate is it what's your confidence level does is is there a way of you doing this that can increase your confidence level with longitudinal measurements and of the same patient talk to me about confidence level and accuracy yeah so we did um we eventually when i when i went when i went back to the doctors and told them that I actually was able to predict the delivery date.

39:49I thought they were going to be like, yes, all right, this is great. This is going to change everything. No, that's actually not what they said. What they said was, you must have done something wrong because again, we told you we could look at these images and if it were possible, we'd be doing it. So we don't know what you did wrong, but you've done something wrong here. So I had to come up, because Carl Sagan said, but extraordinary claims require extraordinary evidence. And so I basically hooked up with Evers Kentucky. I found some researchers there who were much more open-minded about the possibility.

40:21You know, Kentucky has a significant problem with premature births. And they were a little more open-minded to try to find solutions, even if they seemed a little more radical than what you might ordinately go along. And so, yeah, so basically we began a study with them. We got a large volume of data from them because they are a large hospital system there. And they held back 20 % of the answers of when babies were delivered. And then I trained an updated model on using this much larger data set. And then we managed predictions on the ones they held back. And then we did that over and over again for four iterations of the AI because AI was advancing in general.

41:03I was incorporating the latest advances in AI in general. And then, of course, I was learning things, too, on this specific problem, how to improve accuracy. And so we did this over and over again for four versions. And then we eventually published the pair study in the journal Maternal Fetal Medicine and Neonatology, I believe it is. I'm trying to think of the exact name. It's kind of a long name. So it's called the Black Journal in this field. But anyhow, so the pair study proves that you can, in fact, do this. And we got an R squared of 0.92 for predicting days until delivery. So it's extremely good.

41:33And for premature births or births previous to 37 weeks, including miscarriages, we got an R-squared of 0.72, which is shocking. That's on an independent study-by-study basis, not longitudinally. But I think there's an even more fundamentally important thing here that surprises people is there's a concept of an iatrogenic delivery where the doctor has to induce early to save the life of a mother. For example, she might get life-threatening high blood pressure and the baby's got to come out. Even if that's 32 weeks, it's got to happen or they'll both die. And so our squared on predicting when a doctor is going to have to intervene to save the life of the mother is 0.74, believe it or not.

42:15And so that implies a much bigger thing, I think, than just predicting delivery dates of babies. because now what that's saying is that it's possible for an AI to predict a future health condition of the mother and approximately when it's going to happen because it would have to predict that high blood pressure. You'd have to predict when that high blood pressure is going to happen and when it's become life-threatening in order to predict that delivery date because that's when the doctor is doing the deliveries based on that event. And so basically what that shows is that it's now possible to predict future medical events in general, not just delivery dates.

42:49And so this technology actually has a much bigger, you know, possibility of use than just delivery days. You know, we're thinking it could predict, you know, ovarian cancer and all kinds of things. We have to, you know, obviously do the research to prove all what it can and cannot do. But it does, you know, reveal the possibility that, you know, it can do much more and predict, you know, when a health event is going to happen. So it's I think the game has kind of gotten a lot bigger here very recently. but it's just going to take a lot of research and a lot of input from a lot of clinicians to help really figure out how can this be used, how accurate is it, what do we do differently now?

43:29And that's where we're trying to find, you know, this whole new world has been opened up in medicine now, in predictive medicine, you know, and I think there's enormous opportunities but we're just in the very earliest days of this. With a lot of the media stuff that we create with our newsletter, Health Set Pigeon, and and you know this podcast and a lot of different places that that we turn up in the health tech space we talk about women's health a lot very intentionally because of what you'll be i imagine very well aware of a clear data gap and lack thereof of data in women's health more broadly how easy was it for you to train this algorithm and find the data and on what what what data was this trained on how much data was this trained on and particularly like the demographics as well because this is also interesting and relevant that I can imagine knowing a little bit about the statistics of um uh you know early early death and neonatal death and that kind of thing that that you know certain demographics are affected more than others as well and that those happen to be the demographics that we often struggle to get data on and things like that so what was the journey like of you discovering this and going through that to acquire the data and to train the algorithm and the principles that you're i guess operating this company within that I imagine to solve this problem very broadly and particularly to those that need it most?

45:00Oh, yeah, these are great questions. So for the first question on obtaining medical data is really hard, really, really, really hard. It's a significant moat for tech companies to just get it. I can't even express that's how many medical startups go to die in the try to get data phase. It's just, I mean, every hospital system, has some startup founder like, can I get some cancer MRIs? They don't have the time or the money to try to get you your data set so you can go try and see if you can do something with it. And so you've got to have a champion within that institution who's got a lot of pull in the institution.

45:43And there's got to be a convincing case that if this does pan out, that there's a huge benefit and this is going to be a significant discovery that will get our university or system recognition for it. So you've got to sell them on the upside to them. I was going to say, is that the upside that you sell them on? Is that the value proposition that you sell them on, like recognition and research and name out there? Or do you have to explore any financial stuff there? Well, it depends on what the kind of institution is. If you're after a for-profit institution, like University of Kentucky is a big research institution funded by the state.

46:19So I mean, their, their goal is to, to advance research and, and, and, and do, do all that kind of stuff. So that, I think that was more important to them because they're not a for, necessarily a for-profit, you know, system. Of course, if you're going to some other hospital chain that is a for-profit system, they're, they're not going to care about the research. Like how, you know, what, you know, what, how much, how much equity are you giving us? Cause I had many, many hospital systems just wanted to, I had one, wanted half, half the company just to give me data. I kid you not. So, uh, it's, so that, that's where you're, I mean, you're, don't even be surprised if, if they ask for at least a third.

46:53Um, so you even just, just give you data just to try and you've already given away a third of your company. Um, and that's if you, you need to get them to agree to that. Like that, like most of them won't even talk to you. Like even getting your foot in the door is, is, is an impossible challenge. So yeah, so this, that's, that, that's a big deal on the, um, on the second thing is you were kind of going into the concept of bias and populations. And so, for example, black women are much more likely to have a premature birth, you know, than, say, Asian woman or Caucasian woman. And one of the beauties of the AI that I created is since only ultrasound images are going into it, I'm not, the AI doesn't know the patient's race or anything about the patient, all that could bias it in making this decision.

47:37All it has is what it can actually see, just facts, just the facts of what's in the image. And so the AI is actually unbiased completely because it doesn't know what the race of the person in this ultrasound image is. It doesn't know any history about them that might bias a human doctor as to what to conclude. And there's a lot of research that's done on how medical decisions are biased based on the doctors knowing things about the patient that may not necessarily be, you know, or territorially related to the, whatever they're, they're trying to evaluate. And so, yeah, our, our, our representative of the population where that data was collected essentially in that, in that case, it was actually trained on, on at risk women mostly because Evers Kentucky has, you know, they, they take a lot of the, the worst cases, you know, the higher risk women, a lot of the poor women, women who, you know, are of substance abuse issues, mental health issues.

48:36They, they, they take a lot of these cases more so than the average hospital. So this, this AI was actually trained on, on, on a much higher risk population that you'd probably not see elsewhere. And so, so it, so substantially was built on, on women who typically might be biased against. I mean, that's a biased against is not, not the right word, but unintentional bias can creep in, I guess, you know, I don't think anyone's intentionally doing It's just, it's just, that's just, I guess, human nature unintentionally doing it. But anyhow, so yeah, so the fact that we're doing only images eliminates so many ways that bias can creep in, both intentionally, unintentionally, whatever the case may be.

49:19I mean one example might be intentional is you know since a black woman is more likely to have a premature birth then a doctor might in their mind give her a higher probability of that just because she is likely have a higher probability of it but that but in his mind that in her mind that could be a higher probably than what's reality because you know that so it's so it's kind of like that they're kind of using their best judgment you know given you have a historical context as to what's going on but But sometimes that's fallible, I guess, obviously, because we're all humans and we can't make perfect judgment calls every time.

49:52Of course. So that's what an AI is for, is to give you a consistent prediction for every woman, no matter what. No matter how rich or how poor you are, whether you're black or white or whatever, the AI is going to give you an exact, the same prediction, no matter what, without any bias towards your situation, maybe. So that's a good thing, I think. It is. so the other thing that i want to talk about now actually is this the the contention between uh you and this discovery with the clinical world i think that's fascinating because i as a clinician at heart you know as trained as worked for five years you know five years medical school five years as a doctor it's interesting because when um when when someone and I'm going to be I'm going to be frank because why not when someone that isn't isn't medical comes to us and says hey I've discovered this thing that you didn't know it's about a field that you have been doing your entire life and this is a brand new discovery you didn't even think to look here but guess what I've come and found that is so confronting that even when you said it in this podcast I'm like my back is up being like okay what's the evidence here like I'm going to try and find you out here.

51:11It's so funny because even me as a tech evangelist and so open to new ideas and so understanding that there's data that exists that we don't even measure right now. Even for someone like me, I've now got my haunches up and I'm like, right, let's really interrogate you. So I can completely understand and appreciate that the clinical world really struggles with that. There's also, you know, there's also like hardened institutions like built around, you know, this way of thinking being the right way. There's, you know, medical practices that are built around this being the right way that you, you know, potentially threaten for the overall good of the patient.

51:55But what I do also want to acknowledge here is I think because of AI, and this is a perfect highlight and example of it. and I've talked to multiple companies recently, Neco Health being one of them recently, that I think that one thing that AI is enabling us to do is to create new sources of data that we can find correlations in. And I think as the clinical world, we have to accept that there are things that previously we just weren't able to measure that have now become measurable and therefore we can start plotting differences and finding correlations in those things and and that is what we may previously have attributed to intuition or something doesn't feel right or i think it's this because i've been doing this for 40 years i can't put my finger on it but this one is this because of this i think those kind of imperceptible intuitive things are slowly being turned into actual data sources by various companies and i'd almost put this in that category of i think as the clinical world we have to be open to this and it has to undergo the the rigorous evidence and and what you're describing in order to prove whether it's right or wrong but to assume that we know everything about medicine as it stands currently is foolish to assume that we're not going to discover new biomarkers is foolish to assume that we're not going to see more in imagery with ai than we do with the human eye is foolish that is thought that like i i'm happy staking my name to those things as definitely foolish therefore with those things in mind we have to keep an open mind to what is the discoverable however i can also appreciate that it's very threatening it's very confronting for for people with established practices that that have been researching and doing this stuff for decades it's threatening and so i'm interested in how you how you build that relationship and trust with the clinical world and how you interface with the clinical world?

54:18And what has been your experience of butting up to the clinical world with quite confronting information as a non-medic, frankly? Yeah. Actually, no, I wouldn't think it was confrontational. I mean, in the early days, I felt it was because I didn't understand a lot about medicine and I felt confrontational, but it wasn't. What it was, was doctors are on the hook for the actions they perform and then the outcomes that come of those actions. Those lives are in their hands and they need to do things that are only based on solid evidence and not intuition and not, oh, you know, some guy in his basement came up with this idea and why don't I just try it on my patients, right?

55:04No, I mean, I've been challenged really heavily over the years by clinicians and that needed to happen. I needed to be challenged every possible way imaginable because this is this you know human lives are at stake here and so we we absolutely have to get it right there's no room for error and so i greatly appreciate all the doctors that over the years um who who challenged every tiny little last detail because every little last tiny detail needed to be challenged and they're still challenging things and because things still need to be challenged um you know how how well does this work at you know, six weeks, you know, how well does it work on, on predicting this, this, you know, blood, you know, high, you know, blood pressure situation?

55:50What, what, what, what is the underlying mechanism so we can understand how to then develop treatments for it? You know, a million questions and, and we need more evidence about this and we need to dig in deeper into this and we need to, and they're, they're not, they're not trying to tell me my stuff doesn't work or, or that they're not happy that it worked or that, that, you know, that, that they've been doing things differently all these years because they've been operating with the best scientific information they had that, that had the evidence to back it up. And so, so, so they want, I think they want, I mean, I'm certain they all want me to succeed with this.

56:22They just, they just need me to create the evidence and establish the evidence so that they can use this on their patients where lives are at stake and I can fully appreciate it. And I, and I, and we're on the same team. Uh, I just, I just want to clarify that is I don't consider myself at odds with, with clinicians or anyone in the medical community, we're on the same team. And I think we need to challenge each other because I challenge them to look at this, look at everything in a new way. I challenge them and I think they appreciate being challenged. And then in return, they challenged me on, well, now we need evidence, we need hard evidence and we need to do this, look at this very, very thoroughly before we begin using this on humans.

57:01And I think that that back and forth is what's necessary to really advance medicine um that's great to hear and thank you for sharing that so that brings us nicely onto um you were kind enough to email ahead to actually say that we could now talk about this that the the you have de novo fda for this so let's be really specific here around like what it is, what are the claims that you can now make, and what does this now mean for going forwards? What does the plan look like from here for commercialization and what you hope to achieve? I think in the short term, let's zoom in on that bit. But first of all, let's talk about the FDA.

57:49Sure, yeah. And just to clarify too, we are also in South America. We did get an Anvisa approval a long time, maybe a year and a half ago. So we are improved in other countries who allow us to use this in different scopes than the United States does. So how you can use this depends on which country you're in. So I can be the US specific, but the parameters in Brazil are much wider than the parameters in the United States, just to be clear. So when I talk about what it can do, I may not necessarily mean the United States. I'm trying to do this globally, not US century. Anyhow, but yeah, so the FDA de novo.

58:25Wow. So that was incredibly difficult to make. Famously easy process. Yes. Well, I mean, most people don't understand all the distinctions between the different processes the FDA has. One is a 510K, which is you basically, there's another device similar to yours on the market. And you're going to show some evidence that your device is equivalent to this already existing device. and it's i mean it's i wouldn't say it's an easy process but it's probably but it's hey they did something before i do something similar can you approve me and you just have enough proof that yours is similar enough to theirs more more or less uh and to simplify it all now and they grant thousands of those probably four thousand a year of those now de novo a de novo is a technology that there is nothing similar ever to it and you need to create a whole new device category category devices, right?

59:15And they grant about 40 of those a year. Those are extremely hard to get because you got to prove everything from the ground up, the safety and effectiveness. What are the ways the device could be misused? How could the results be misunderstood by a doctor? If the results are misunderstood by a doctor, what could happen to the patient? How do we label this properly so the doctors understand how to use it properly? It's just everything. And so a lot of large companies have struggle with a de novo approval because it's just such a monumental endeavor and for a tiny little startup some guy in his basement getting one is just uh it blows people's minds like disadvantage they probably call that yeah yeah i mean but it's uh but you you have no other choice right so anyhow so we defined a new so we had a new product category in our category code and i maybe it's it's possible people the fda have a small sense of humor here, but they gave us the code of a S H E for this device.

1:00:17And generally, uh, these, these product codes are, they appear to be just random, three let random letters like X G Q or something. And, uh, I, so I, and the fact that this was S H E for a, for a, you know, delivery device, I, I imagine, I can't imagine that was random. And I, I thought, I thought maybe they were just having a little fun there at the SDA. There's nothing wrong with it. It's a good code. I love it. Cause you know, there's a little bit of marketing, uh, you know, you can just say we got the code and it's, you know, she, right. But, uh, so yeah, so getting the, the de novo, but now that we have the de novo, we are able to now expand upon the use.

1:00:47And let me, let me describe the exact parameters of the use in the United States that they've approved because that, that's very clear. Um, basically it's, it's for a high risk population that doesn't have prenatal care. So it's, uh, women who have not seen a doctor, you know, before 14 weeks at all, She may not have any idea how far along she is. So she's showing up to the doctors at 14 weeks or later, not knowing how far along she is. And the doctor, the best way they can tell is just by doing some measurements on the baby. But those measurements are pretty inaccurate. And it's quite common that if someone doesn't have prenatal care, then they may have substance abuse issues, malnutrition.

1:01:27They're likely to be quite poor. And the baby's measures be off because they're typically smaller than the average. And so that's kind of an inaccurate way to date the pregnancy. And so the FDA agreed that, yeah, for this population, this device would be effective and useful for them because it's better than what the doctors have now for that. And so basically the limitations are you have to be greater than 18 years old. you have to be at least 14 weeks long and not know how far along you are in the pregnancy or just not very confident in that number. It's not for anyone with fetal anomalies. There might be some other minor, but that's the major indications for use that actually are critical to follow.

1:02:18But obviously, we have our device details, indications for use when people we're going to use it. But, you know, so, but we're now we can file 510 Ks to then convince the FDA that, okay, now, now let's try to prove that it's, it's safe for as early as six weeks now. And so we gotta, we gotta come up with that evidence, submit it to them, file for that, and just, you know, do all the process again, should be a little easier. And even though the FDA process was extremely painful, I just want to say that, I mean, these people are dedicated. The people at the FDA are very dedicated to making sure that what's released to the market is safe and effective.

1:02:57I'm sure at times, because I'm not a startup founder, I want to move faster. And I'm sure at times I thought, well, is this really necessary? Is this really necessary? But it may be. It may not have moved as fast as I liked, but it's their job to make sure that they're releasing safe and effective devices to the market. And I think they're doing their job well i i think they they are dedicated people who who care deeply that that that happens so also thanks for saying that robert so i can't remember whether we're recording or not when i said this to you at the beginning but the the this is this is part of my thesis here like increasing the amount of empathy we have for each other in this ecosystem because the more we understand each other the faster information can actually flow the better the information flow will be and therefore the better place will end up in in health tech i think this is that is a perfect example of it that yes you're on opposite sides of the table there but you you are ultimately both on the same team let's get a safe device out to patients that does what it says it does that's ultimately the goal that you both have one of you is pushing the other one is doing the checks and balances but i think it's a really nice thing you just said i think the empathy for that has hasn't gone past me i think that's a really important thing to to to actually have and and to communicate i yeah it's the same as the doctors right it's the same as doctors they challenged me to prove everything that i was saying because human lives are at stake and so like the doctors challenged me on the we need evidence of this we need evidence of this we need evidence of this they have to be basically in that role too but as the government's role of that you know and so it was it was the same back and forth but now having gone through that that that rigorous scrutiny um i think you know i feel much better about what what i've released to the public now because it has gone through so much so much scrutiny i can feel better about it myself too so so you know that's i think it's i think in all it's a good thing because it's all that matters is the at the end of the day it's the patient right so yeah do you take time to reflect on it you know when something like that happens do you take time to think about what you've done from those early days and actually now there's there's there's people experiencing this that wouldn't have otherwise there's there's there's impact been made oh yeah south america we've been in use for quite a while.

1:05:13So in real world clinical use in quite a while, I mean, we just got approved by the FDA recently. So we haven't had any real clinical use in the United States yet, obviously, but, but in South America we have, and I've got, uh, some amazing testimonials from doctors who just swear by this. And, you know, some of the stories they, they, I have are just a mind blowing and, and it just, you just want to keep, uh, doing more and fighting harder and, you know, seeing how you can expand this to help even more people because it's just so encouraging that, that there's real lives who are being dramatically, you know, dramatic, like, like you're saving the lives of their babies.

1:05:48And, uh, you know, so it's like, it's, it's just so, it's just so powerful and it's really so motivating. And, uh, and, you know, I never really, uh, until recently I didn't have that motivation cause it really wasn't being in use. I was just hoping one day that would, I would have that motivation. But now that I do, it's, it's just extremely, extremely, I can't, I can't stop now. There's no stopping now, no matter what. Well, this is what's exciting, right? So you've got now a commercial set of challenges in the US to make impact and the level of impact that you want. You've also got R &D challenges, because as we said, you're at the frontier now of a lot of potential new discovery.

1:06:27how do you think about the growth of the company now in terms of like capitalization of the company in order to in order to actually give you the fuel and the resource to attack these different things like how are you how are you thinking of balancing you know revenue versus actually raise and put it into r &d and give you more chance over here like there's there's there's lots of different like priorities you could have here in terms of how you go forward so i'm interested in in how you think about this now? Well, yeah, now that we have the de novo, I mean, money was a struggle to come by before we had approval.

1:07:02I raised a total of$8 million. By the way, I'm curious how I raised money for USA. Basically, safe notes from a credit investor, basically. But safe notes. So money was kind of just trickling in. I had to make it work with that. So obviously$8 million is not a lot of money to do it. That's incredibly impressive. To do it de novo and actually have all the patents that we have and all that. But yeah, so now we got the de novo and it works and everyone's like, oh, wow, this is real. Now we can sell it to the United States. And then FDA approval is more or less approval in much of the world too, because a lot of countries just accept FDA approval as is to some degree.

1:07:44And so now money's not so hard to come by now. And we're beginning to work on, we're doing a bridge round right now, which is pretty, you know, our current investors are, are, you know, kicking in more cause you know, they're, they're pretty pumped right now, as you can imagine. So anyhow, but we were working on our series a round too as we do that. And we're getting a lot of interest from the OEMs, the you know, the companies that make these ultrasound machines, you know, they're, they're interested in embedding this directly into the machine. So instead of it going to the cloud for prediction, the doctor can have it directly on the machine.

1:08:16Of course, there's other the insurance companies are kind of really excited about this because they're the ones on the hook for the 30 billion dollars a year in NICU costs and so of course it'd be cheaper for them to pay for every woman to have this if it can prevent a certain percentage of the cost you know 30 billion dollars even a small percentage of that is an enormous fortune right and so they're on the hook for the money and of course mothers um you know want want to just know this date for to relieve their anxiety so they know everything's okay and also you know if there is a problem that it's taken care of.

1:08:46And also for vanity, sometime of, well, should my husband go on a business trip around this time or not? Or there's that angle too, which is worth something too. So some patients are worth paying money just to get this knowledge even before it's covered. So there's a lot of angles for generating revenue here. But I just want to clarify though, that my goal from day one has to make sure that this is available to every woman on earth, regardless of their ability to pay. So if someone in Zambia or pick a country can't afford to pay for it, then that's fine. It's free for them. And that will be worldwide.

1:09:25So I think that's the promise of AI that we've been selling the world for the last 30 years is that eventually this will be so cheap and everyone will be able to have everything they need. And so, well, now it's time to deliver on that promise, right? So all the AI people who've been making predictions over the last few decades have always promised this. And so it's time to start delivering on that promise that AI is here to create. And I'm happy to lead the way on starting that, delivering on what we promised forever here. Amazing. Is there a particular country or particular area or region that you're excited about?

1:10:01It depends on what the angle is, I suppose. I mean, I'm really excited about the countries that really have no maternal care at all. Because ultrasounds are getting really cheap. The machines themselves are getting really cheap. And there's a lot of nonprofits around the world. There's a lot of point of care ultrasounds getting. Yes. Canthels. Semiconductor stuff. Yeah. They're getting very powerful and small and cheap. So sending a box of them to a poor country, a poor village is actually very realistic now. So now we need the, but they don't have a, they don't have an obstetrician in their village to actually read these, these images.

1:10:35Right. So, so what I'm really excited about is getting AI into these handheld devices in these remote villages so that these women can actually have a high level of care. Yes. That would have been completely impossible, you know, until now. So I think that's also the promise of AI is getting knowledge to places where knowledgeable humans, you know, knowledgeable humans aren't scalable. Right. but AI, you can make copies endlessly of that. And so I'm really excited about how we can deliver healthcare to women who really don't have it at all or very little of it and start delivering really powerful capabilities that will save their lives who would have not had any chance before.

1:11:17So I'm excited about that. Amazing. Are you doing anything in the UK yet? uh we're we're working we're we're this all the uh all the developed countries are a little bit slower process because uh i mean fda is great but uh obviously the but uh the you know japan and you know canada they don't just you know oh the fda proved okay we're good i mean all these all these developed world head countries have their own ideas of what's safe and effective and you know they have their own standards so that's we're trying we're trying to get into the developed world as quickly as possible too but it's it's a lot faster to get into the the underdeveloped countries because they're generally more accepting of FDA approval.

1:11:55Okay, yeah, you're good. Go with it. Let's go. You're welcome here. So, I mean, we're trying to move quickly to cover as many people as possible. The underdeveloped countries are much faster moving. And since this is a cloud based system, you can scale pretty rapidly. But with the developed world, there's a lot of regulatory work that still needs to be done because they all have their own ideas on what's safe and effective. We just got to do their processes now. But having FDA approval helps dramatically because we've got a big, solid amount of evidence that we've already created that we can just repackage into whatever they need.

1:12:30So we're moving as fast as we can, but it's just a matter of now we got to raise a lot of money to hire more people to do the work. And so right now, scaling up is our biggest problem. Everyone's calling us. Everyone wants to do this. Everyone wants to use this. And we're just like this little startup now trying to, you know, expand explosively. And that's by far our biggest, our biggest problem right now is it's, it's hard to scale up a medical type company explosively. It's, you know, so that's, that's where we're struggling right now is with the explosive growth. I mean, it's a classy problem to have, but it is still a problem.

1:13:03It is. It's a, it's a classy problem to feel that you have as well. The, the, the whole, I don't even like the term, but, but the whole, you know, the blitz scaling term that got thrown around, you know, a fair few years ago, you know, it's all, it's, it's, it's never felt comfortable in healthcare just because of everything we've talked about in this podcast. And so, yeah, it's, despite there's been a few people that have tried, there's been a couple of companies that have certainly tried definitely in the UK that famously, and actually a little bit in the US that famously, relatively recently, that didn't end so well.

1:13:44I won't mention any names of you, Doug. But no, it's nice to know that you have those principles as well. And Robert, honestly, it's been an absolute pleasure. Thank you so much for telling that story and what a story it is. I wish you all the best. I hope that all the R &D that you're doing proves that the evidence is there for more and more and more of this. I remain very open-minded that there is so much that we currently don't know and don't see that I hope technology will prove to be more good than it is bad. And I think certainly in healthcare, there's plenty of spaces like this where we can learn a great deal, pick up far more biomarkers and start getting so much better at preventative healthcare and helping mothers through their pregnancy journey and helping those neonates get the best care that they need.

1:14:47For people that want to learn more, Robert, what's the best place for them to go in order to do so? and by the way just to respond to what you said i really appreciate you helping me get the get the message out that this is now possible and and the all the potential here and all that that was i mean it's you got to make this people aware that this uh is possible and then and i would like any anyone that's interested in doing research or working with me or helping refine all this uh you know please please reach out to me um and also you can if you want to find out more i mean i can be reached at bob at ultrasound.ai if you want to reach out to me personally to do a collaboration or research or you know you want to know more feel free i'm i'm a founder that's very passionate about this so i'm happy to talk to pretty much anyone um but if you want to just read more about us in general of course you can you know look up our papers online like the pair study from yours kentucky but just our website is probably a good starting point because that's a good source of links to all this other stuff and that's just ultrasound.ai you know www.ultrasound.ai so i We got an amazing URL, obviously, here.

1:15:49I was going to say, what a fantastic domain that is, obviously. I haven't even talked about how you acquired that, but maybe that's a different podcast. Yeah, possibly. It's a good story. Absolute pleasure. Thanks so much for joining me. Thank you. I really appreciate it. You have a great day.