Welcome to Real Talk on AI and I care with your host, Dr. Scott Morris. And I'm co host, Dr. Rehan Ahmed.
In each episode, we discuss current news, adding a little AI education, and debate innovative AI tools and topics that will change the industry. All right, everyone. Welcome back to AI and I care again. I'm Rehan here to top story of this week.
So top-con healthcare and seems to Scott, they've been in the news a lot and making investments left and right. Seems like every week we hear about something. Every week they've been busy and they're really leading the way. I think a lot of this stuff and great things.
I think it's it's the rising tide list list all of us. So I'm, you know, it's great that they're doing. They made a huge investment in company called, you know, I didn't know about this company before. I just showed you just how many companies there are that are new, but doing very, very well.
It's called Pangaea data. This is a health AI company that's specialized in. And they're sort of niche is identifying untreated and under treated patients that get so big. You know, you know, from obviously an ophthalmology and optometrium, we ask our notes, but in hospital notes and there.
So there's so much unstructured data patient complaints, you have lab reports over hospital visits clinic visits. There are a lot of patients who probably who definitely have diseases or conditions that have never been diagnosed or a misdiagnosed. So Pangaea has a platform they call it palates P-A-L-L-U-X. And it sort of emulates clinicians review by automatically analyzing sort of structured.
And here's the unstructured patient record. So when I say unstructured, it sort of means it's all the random text that's in the in the history and the chief complaint and assessment plan. That's that's basically unstructured text data. There's a lot of information there that we just we don't even know is out there because we've never really had a way to collect it or connect it before.
Right. Yeah. Yeah. And and and books structure data is kind of like your lab report structure in the easy to you know in a table or something, but there's a lot of unstructured records and a lot of energy too.
And this so the idea here is that palates which is a platform and it integrates allows an integration for real time and flex without distracting work flow at all. So the idea and they have released a ton of information, but I think the idea here is that Pangaea is the AI platform will for be built in to top known healthcare, you know, their harmony digital platform and the whole health care from the I initiative. And the idea is for proxies to identify patients who need follow up or further value. So an idea that could be some like well, we have a ton of patients who are undiagnosed.
I mean, they just is a lot of unstructured data from imaging that you know made me they've been undiagnosed or some type of inherited retinol disease or diabetic or not to be where it's just the data is there, but they've never been diagnosed. So when they start combining that with machine vision, being able to read no CT what that how that might transform clinical diagnosis. I mean, all that unstructured data that we just take as little bits and pieces that maybe we discount here and there, but then you start you start collaborating that with OCT data that's on machine vision. I mean, oh my gosh, we might change the way we diagnose lots of conditions.
Yeah, you know exactly right. And so I really think this is in line with the health care from the I initiatives that it is the major initiative and we'll be talking about Oklahoma here in a few minutes. And so I'm really excited about this because and I've talked about this a lot the eyes are, you know, as we sort of learned school, the windows to the body, right. And this connects us and I think top on again is really helping us lead the way on this and connect us to the rest of health care in terms of other conditions and diagnoses and showing how I dare and really is health care.
And so really excited about this to see where this goes. So that's the news of the week. It's always so staffing how there's something or something new every week in our niche field. I think you brought up a good point at the beginning, Rayhan is I think about how many companies are out there now in the I world.
Then I think about, you know, a couple years ago, I built a big grid that says here's all the companies that in some way touch I care. And then I started building the same grid. In fact, I'm going to present it at Vision Expo West during our AI and I care summit meeting all the different companies are in AI. And you know what, they're almost equivocal at this point in terms of how many AI companies are coming out of the woodwork compared to how many companies in traditional optometry and ophthalmology are out there.
It's quite, it's quite stunning, you know, in terms of how fast this little subclinical space is growing. Well, today in the knowledge by it, Rayhan, I wanted I was having this conversation this weekend with a couple of people who are friends of mine. And they said, it's got explaining me what generative AI is. I don't really get.
And I think it was key. They said, what's G.A.I. And I said, well, we've got to break that down. And we talk in generative AI.
And we're talking about general AI. And we're going to focus on generative AI today because I think a lot of people go, oh, I use Chatchy BT and you know, I'm using generative AI. And I'm like, do you really understand what it means? And so it made me think about maybe that's a good topic for the day for our listeners in our what we call our knowledge bite.
So I want you to imagine a world where computers not only understand information, but also create something entirely new. Now, generative AI is not new. I mean, it's been around for 20 years. It's only now that starting to hit let's call it the innovative mainstream where people go, oh, I can create some captivating story or I can develop a piece of artwork or maybe I can write music with it or even and more importantly functional line of code.
I was having this discussion with my business party today. I'm like, how much code can it write and they said unlimited. Well, this is kind of the power of generative artificial intelligence or general of AI, which is really changing pretty rapidly how we interact as humans with information and what the realm of creativity is and maybe could be. So what exactly is generative AI?
Well, I mean, it's a type of AI that we've talked about. It can generate new novel content and unlike maybe more traditionally, I systems that might analyze and categorize data. Generative AI actually produces new data kind of creative like our own brain does. It learns from vast amounts of existing existing data, whether that be text or images or music.
And then it uses that knowledge and sometimes various types of that new knowledge to kind of create new and somewhat original outputs, which we might call a story or music as I said. Well, how does this all work? Well, think of it like, well, maybe a highly sophisticated student if we think about we study lots of examples these AI muddies, these A and model study lots of examples. They take these complex architectures we called neural networks, which we learned about a couple weeks ago.
And it says, let me learn about these underlying patterns structures of how music good music is created and how good art is created. There's always patterns and structures. And so really generative AI is not necessarily creative like is building something totally new, a totally different pattern structure that ever existed before. But it instead is saying, what are the underlying patterns and how do I take some content, transforming into those patterns and create new content that follows those patterns and structures.
So when we think about using some prompt or instruction on GBT or Gemini or, you know, Copilot or whatever, we're basically giving them instructions to draw upon these patterns and structures to generate something new that kind of aligns with whatever we thought about. You know, it's still original, but it's using existing patterns. And we're starting to see that generative AI actually all around us. I ran out of this thing about their days like almost every now.
I'm getting to the point where I can look at an article or look at something that's turning it. I go, I can tell if that's generative AI or not, because generative AI typically follows a pattern of speech. Whereas if you have somebody writing it, they many times will jump from first person to third person and maybe the punctuation is always perfect. So I'm starting to read it going, man, there's a lot of things being written by generative AI.
And maybe that's chat bots, you know, holding conversations. I was on the phone with my mortgage company this morning. And it was definitely a chat bot for the first four minutes. I said, Hey, how's it going in the answer back?
And I said, nice weather. And this is a cold in Seattle. That's where I'm at and the computer got lost because it knew I wasn't really calling from Seattle. You know, so it I was just trying to trick it.
But you know, I think there's my marketing people, they use these image generators that makes avatars and we feed it script, we feed it text and then it turns into these crazy amazing audio visual types things. I mean, I I use it all the time to write emails or generate code when I'm building something. I think gender AI though it's really far reaching covers a lot of things. It really does have the potential.
And I know I use it literally every single day. I know we've talked about, you know, I use Fred as my AI assistant that's kind of my scribe. But I also started to use I was we talked about one of the earlier podcasts. I love Gemini.
So, you know, I I plug something and go, well, let's think about that. Let's like, how do I automate tasks or how do I be creative and think of a new way to visualize things and a new way to say things. I think generally, if I represent really this huge leap, Brown in terms of yes, it's artificial intelligence, but it makes us as humans think differently, process differently. And maybe we'll allow us more time to create new things and think about things differently and maybe open up a world of possibilities.
We are just starting to understand. Oh, it is no question that we are in really thinking. And I think we have talked earlier about how is a great type of a question that one of the founders of Google, I think right there on the top of that AI is actually under high, not not over high. We're definitely really in the gen A.
I mean, gen A is what capture the imagination of everyone. And so that's, that was a great knowledge, but just get everyone up to speed on what on what gen AI is for the fireside cat. I wanted to talk a little bit about ocular looks. And this sort of connects to what we talked about earlier in terms of top on investment, Penti help and their healthcare from the I initiative.
So the idea behind I go up and as a term that's been very popular in the AI and I care circles, so definitely want to make sure our listeners are up to speed with that. I've been on this as a phrase that was going into just a few years ago, there was a bit published in from London from more fields. And basic idea is not, but the idea is not a new idea when I remember I was in school 20 years ago now, like well, actually, but we had a plot called the eye of the sense and a lot of disease. And so there's a bit classic stuff like, you know, the Kaiser fly sure rings that we remember.
Yeah. And welcome to the sea, but for like, discoloration of this square, the indication of cirrhosis or liver disease, or of course, vascular changes in the retina to show hypertensive right now. So it's not new. I mean, one of the powers of I care that what looking at someone's eye, we have a study.
Yeah, I don't know how to do. Not just the windows to the body, but windows to the soul or vice versa. Yeah. Right.
So there's something new that I'm and the new thing is when if you'd ask me when I was in the next school, thing that class, if you showed me a picture of someone's retina say, hey, rat, is that written a belong to a man or a woman. I'm going to look at you a lot. I have no idea and no one has any idea that the ridiculous question we don't know a couple years ago, Google published a paper where they show us 95% accuracy, actively categorizing a man or woman based off a rental photograph alone. Isn't that crazy crazy.
And the reason we're up, though, is interesting because it finds patterns that are not entirely non obvious. And then we have to go back and serve back fill a rationale for why that occurred. Well, is it or why we didn't see it or why we didn't see it in. Is there something maybe a background segmentation?
Is there something in the best your size of the vessel. I don't know. You may I don't know if they've actually figured it out. I don't think they have.
I don't think there's a window. Some people say yes, some people are saying it's vascular diameter of the Corridor vessels. Some people are saying it's pigmentation. I don't know the answer to that, but I think it's fascinating.
We were I was having this conversation with some of their day about glaucoma. You know, and said that I think in the next five to 10 years with what machine vision is going to do, it's going to be a little look at blood flow patterns by heartbeat and see what blood flow is through the optic nerve. And we may find that we've been treating glaucoma all along because glaucoma might just be a vascular perfusion issue, which has much more to do with the whole body than just the optic nerve. Yeah, yeah, the good evidence of that from looking at you know patients have a lot of pressure friends.
I definitely see something there. And that is that example. But and so that's just like so the weather manner that's kind of the interesting sort of parlor games, but what's start thinking about that? Well, what does that mean about cardiology and cardiovascular disease and nervous.
Well, Alzheimer's, right? I mean, they can nerves of general disease like Alzheimer's or Parkinson's or MS. I mean, you know, if Alzheimer is really a vascular abnormality in cerebral blood flow, which is what some people predicted to be, then maybe what we're seeing, maybe there's a relationship between Alzheimer's and AMD. There's been a lot of people saying those two things are somewhat related specifically geographic atrophy.
You know, I mean, gosh, this is a window to the brain, not just windows to the soul. Right. I bet you know, the often see is that the retina is the only tissue in the body that's directly observable. And you can actually look at brain tissue and blood vessels.
So that's where the power is. But you know, there are some challenges here. And I think as evidence of that, there was a NIH had a active moment to issue that they had. And last year and in and in 2024 where they had a whole bunch of very big labs, big and small labs applied for funding, looking at applying novel non-based of eye imaging.
And I think that was the first thing that I thought was when I attended their talk was that there is an upper limit to the to what we're going to get with traditional fundus photography. Absolutely. And it's great. It's fantastic.
And we and we may just be getting there, but there's probably not for one that in terms of fidelity of it's the data, right? Yeah. But there's still kind of proved, especially when you compare them to something like adaptive optics or OCT where you're getting cellular level neural tissue. And so I think we're just sort of getting started with on the long mix and I think fundus imaging retinos.
It's always tied with retinal sort of traditional fundus photography, but it's far beyond that. It still involves retinal in the being in tissue, the right because it's pretty much pays place in our, you know, and so what we described about being the only brain tissue that's observable, but it's far beyond that when you go to the cellular level. And so really excited about seeing what else we did on the public. And I think sensitivity is specificity of really predicting diseases that are years out.
Again, you were just getting started identifying those types specific biomarkers. I think biomarkers, like my big thing is dry eye, I do a lot of dry eye and I will tell you what I mean over 25 years, I've gotten to the point where now I can look at somebody and within a few seconds go, how's your hormone profile? Like let's go find out what it is because I think you probably have a testosterone as a woman. And it's not necessarily a private testosterone deficiency or there's some type of endogenic issue going on or whatever.
But that's just me in 30 years of gut instinct and, you know, seeing a lot of patients. But I think in the near future we're going to be able to analyze a tear film, feed it in a eye and it's going to say, here is instead of blood work, here's your entire hormonal profile as well as any inflammatory markers you have. Not to mention, you know, blood sugar, we didn't even get into that about what the future of being able to read all of this stuff with a contact lens on the surface of the eye. It may be more effective than you're ordering, you know, or your Apple watch.
I think that oculomics is a fascinating what we may find out 24, 7, 365 real time analysis of what's going on from a biochemical profile on our tear film. Yeah, no, exactly. And there's also I just want to get our listeners again at the speed on what acolymics is. There is a new alliance that's we published it in AI and I can make sure everyone to check it out regarding the Alliance for healthcare from the eye.
And there are a number of organizations and companies and I've been in the groups that are looking into this. But again, I think as we're all working on this together, I think it's just just what elevate the status of of eye care and being healthcare. And that to me, I love that I love that breaking because it doesn't isolate or separate us from the rest of the rest of medicine. I think that we can probably do a full like in fact, actually, I know there's a couple of talks coming up at some of the major meetings that are doing two hour talks on acolymics.
And I'm like, oh my gosh, I can't wait to go to those because I think that we talk about it maybe sometimes in a laboratory setting. But then I think the context of which AI will be able to read through volumeist data about of day to day or minute to minute blood flow analysis at the optic nerve or tear film proteins and metabolite. And metabolites and what happens with blood sugar during what you eat and how that transposes into the tear film. I think that when we start feeding AI, this feeding AI, all of this crazy amount of data.
I think we're going to see trends that in the, I don't know how many years eye care has been around between ophthalmology, I'll start with ophthalmology, hundreds of years. I think we'll find out more in a year than we've probably found in 100 prior to that when we have the power of AI to really look at that data and dig in based on my favorite phrase, which is phenotype, genotype. I think that acolymics we will be the windows and I'd say even the other way around it's not just the data we take out. It's the data we take in.
I think you could potentially add in, maybe this is a little bit of stretcher, but if we look at. Head's up display and virtual reality glasses and all that kind of stuff, not as it just the data that's happening inside the body, but what's the data coming into the body and how do we respond to that some of the psychological and psychiatric implications of that are pretty impressive as well. We start with new today that every week is amazing. You've been listening to Real Talk, an AI and eye care, your weekly podcast to keep you informed about AI technologies revolutionizing eye care.