If you can save a patient from ever, you know, going blind from diabetic retinopathy in the first place, then obviously you're going to save the health system tons of money over time and of course the patient. The talent here, the superpower is to find a solution that does really solve a problem, but it also generates ROI for all stakeholders. What excites me most is the growth of precision medicine. Hi, and welcome to the AI Innovators podcast, where we talk to leaders who are shaping my care.
My name is Rahan Ahmed and I'm the co-editor at AI and iCare by Jobssen. I am absolutely delighted today to talk to a good friend and a mentor of mine, Dr. Rania Habash. For those who don't know, Rania is a comprehensive ophthalmologist and a digital health entrepreneur.
She's a voluntary system professor of ophthalmology at the number one rated vast compalmer eyes too. A visionary innovation mentor at Stanford and she's on the FDA's Digital Health Network of Experts. In addition, she co-chairs the AI committee for ACOS and has been named to the ophthalmologist powerless multiple times, including as one of the top 50 stars who will lead ophthalmology over the next few decades and will shape its future. Thank you so much for joining us today, Rahan.
So Dr. Habash, we talked about a little bit about how a person who's new to AI, what sort of tools and technologies they can use to get their feet wet. Now I want to talk a little bit about what are the big opportunities in iCare. Again, thinking about the whole vertical of optometry, ophthalmology, and optical for infertility of iCare.
And really maybe even healthcare generally. So where do you see this going? Where do you see the big issues where AI can help then at iCare. So I would start with what we just talked about, which is the operational and administrative improvements, things like natural language processing for real-time translation, patient education, AI agents like you were saying, and ambient dictation.
I think those are all really important workflow tasks, right? But then beyond that, we have diagnostics. And this is really where I think generative AI shines. And you know this as well, you know, with some of the newer technology like OCT, OCTA, we're able to find things that we would never normally see with the naked eye.
But instead AI systems are then guiding us to look at certain abnormal regions or pathologies that we would never pick up normally. And so those systems are getting way more advanced over time. The digital visualization aspects are getting way more impressive over time. And so those types of diagnostics I think have a long way to improve and a long way to help us even further.
Yeah, it's amazing what even something like chat TPT can do. My mom had blood work done recently. And you know, I'm an ophthalmologist. We did do a year of internal medicine, but you know, interpreting her lab maybe was something I'm not totally comfortable with.
So I just stopped the picture and I uploaded a chat GPT. And it gave me an amazing differential. Oh, wow. That's great.
I haven't tried that yet, but that's actually amazing. I would love to try that. And I want to say, do you, where do you feel like, you mentioned OCT? So I've been a little bit deeper there and maybe the diagnostics in general.
Practically, I'm a non-pumpturist who maybe it has a device, but maybe findings that they're not. Is there like an easy button? Do you imagine an easy button where AI can come in and play a role? Well I think in terms of OCT, for instance, it's the digital visualization aspects of it.
You know, we're able to now find these smaller patterns of anomalies better and more specialized than we ever have before, right? But that's just one bucket. Aside from that, there is, you know, ultrasound biomyroscopy, corneal topography. Imagine for not only finding early, you know, signs of disease, but also predicting disease over time based on the data patterns that are being interpreted.
And then another really important point for a lot of the optometrists and opticals that are out there would include things like AI diagnostics, autonomous diagnostics in fundus imaging, for instance. And that's already there. It will just continue to be improved even further. That's one area where I'm really excited because, you know, snapping picture of the eye, it used to be, it used to be able to see diabetic retinopathy.
Well now you're going to be able to autonomously, when I say autonomously, what I mean is it's just a machine, it's not a physician who's interpreting this. There should be physician oversight of course, but the machine is then flagging or telling you if there are systemic conditions, things like hypertension, chronic kidney disease, cardiovascular disease, biological age. And you know, I'm really excited about things like Alzheimer's, for instance. So you know, those are the types of AI diagnostics that we're going to see in the very near future.
And I think that your audience should take advantage of that for sure. Yeah, I think that's the whole idea of octalomics, right? Where you see the eye, the sentinel of disease and other, we sort of learned about that in medical school, in terms of hypertension and maybe some other neurologic conditions, some zebras that are out there. And that comes from Alzheimer's and neurologic conditions from a fundus photo, especially donuts and opticals in the way where these are patients who may never actually go in for routine eye care.
I think that's really... That's right. So if you're a large optical player, how would you sort of think about the economics of it? I mean, I guess it's a tough question, right?
So do you think about that at all? And how would you...? Of course. So, you know, my bias is, of course, always just to improve access to care for patients.
And so, you know, let's think about something like ROP, which, you know, in certain areas, you know, there isn't a sub-specialist trained pediatric ophthalmologist who's going to come and screen those babies, right? But imagine now snapping a picture instead, right? And then sending it and having that tertiary level care to every corner of the country or world and giving access to care for those patients where there may not be care. Now, if we expand that to things like optical shops, I mean, most towns will have an optical shop.
I grew up in very rural West Virginia, right? We didn't even have an optical shop, but we eventually got one and we didn't have an ophthalmologist though. So, for a lot of those places, I think that the optical shop will then serve as the hub for vision screenings. And then the patient can be referred on to the next level of treatment if there's something that, you know, that's flagged.
But that's where I think AI-powered diagnostics are really going to help our optical and optometric counterparts as well. Yeah. You mentioned growing up in rural West Virginia. I grew up in sort of rural Southern Illinois.
And I can imagine that, you know, a lot of patients may, you know, when they can't see, they go see someone, right? Right. So, you know, undiagnosed hypertension or undiagnosed, you know, early neurologic symptoms and giving them caught early could be a huge advantage. That's right.
And you were asking about ROI and that's where hopefully I was eventually coming. So even if there's no, and there is a reimbursement model anyway for autonomous AI, thanks to the FDA and CMS now. So it's not a lot, but at least it's a little something that can supplement the exam. But really the bigger picture here is not so much the reimbursement.
It's the healthcare savings over time. And that's where I really think that the ROI is going to pan out because if you can save a patient from ever, you know, going blind from diabetic retinopathy in the first place, and obviously you're going to save the health system tons of money over time. And of course the patient, all that agony and harm as well. So I think that's really the next frontier.
That's what I've been really devoting most of my time and attention to is predictive analytics and precision medicine using these AI powered diagnostic tools, but then using AI to start finding patterns within that data and not just retroactively going back to treat a patient, but starting to predict who might end up with a disease over time based on all of their academics like you were saying, or who might benefit from a treatment more than someone else. I'll use your terms in retina for instance, who might benefit from one Begef inhibitor versus another versus treatment extend versus every one month treatment. Those are insights that we can derive pretty easily from the vast amounts of data that we're generating from this diagnostic AI powered diagnostic equipment. And your testing on a lot of topics I think are relevant to the FDA clinical trials, not only retrospectively, but prospectively looking at data and how patients respond.
Would you touch a little bit on how AI and eye care can impact? I know it's a huge topic, but how it can impact clinical trials and drug developers. There are several different ways. In eye care specifically, one of the really interesting things that I work on is using the eye to monitor for other things.
And so what I mean by that is imagine a pharmaceutical company is doing drug studies for a new depression or anxiety medication. You can actually use the eye to gauge that patient's treatment over time and start predicting the patterns and how they might react to that drug. So just using eye movements or pupil size, blink reflex or blink analysis, it tells you a lot about the patient's mental state, for instance. And so using those types of really easy tools from the eye can help inform our pharmaceutical partners in like an it's an RPM or IOT type of methodology where it's remote patient monitoring, which improves adherence to their clinical trials and then gives them much better real world data that's longitudinal.
And so they can really refine the drug based on that. And the big idea is that in eye care, this is like a mask, we talked about clinical trials, we talked about diagnosing conditions, we talked about predicting conditions. I mean, it's touching every sector of eye care from optical.com, pre-doubt, mology. Thank you.
That's right. Talking about big ideas in a very short period of time. Thank you again, look forward to continuing conversation. My pleasure.
Thank you, Dr. Bosch. The next topic we'll be talking about is something that's near and dear to my heart. It's how AI has been impact, it's startup ecosystem, particularly how it relates to eye care.
So you are an entrepreneur, you're an entrepreneur, successful entrepreneur, you're part of a number of different companies, you have a ton of experience. And I know you get exposed to probably a lot of companies that come to you and say, tell us about product market fit, would you see this? And so give us your best practices or what you're looking for or just advice in general for startups in this scenario. Yeah.
You know, a lot of times I'll see founders coming towards me and they may have heard of one small pain point, but they're not really within the field. And so they don't understand the real pain points that we're trying to solve for in medicine. So one thing that I would urge them not to do is to try to solve for one very small thing that really isn't much of a pain point. And in order to make sure you understand that, if you're not within the field is to talk to the key opinion leaders within our field and see what really matters to us.
Now that can be double edged sword sometimes because the problems that we want to solve don't always have an ROI for the industry, right? So I think part of the talent here, the superpower is to find a solution that does really solve a problem, but it also generates ROI for all stakeholders. That's one thing I think I'm pretty good at and that's something that I would urge any startup to do too. One thing I found challenging is our reimbursement mechanism in this country is very difficult to extract ROI on something.
You mentioned that with with fun to talk about it. There is an avenue. That's right. It's gone down, right?
And it's hard to predict what's going to happen over the next three to five years even. And so coming to you that would you advise companies to follow reimbursement pathways or would you think about other potential business models? You mentioned something where there's a population benefit, but that's what's the topic. That's what's the topic.
It's convinced investors to say that there is an ROI. So how do you think about reimbursement for the type of software? Yeah, no, I think that's great. So first of all, there's tons of companies out there that really don't have some reimbursement mechanism or pathway.
And that is already a big red flag. Even if there is, and the AI retinal imaging that we talked about, AI power retinal imaging is one of those things where it does have a small reimbursement, but that's not going to make you get out of bed in the morning. But the population implication for that is massive. And so I would structure that in a way that this is a huge ROI for payers, for instance, or for government plans.
And once you start looking at things from a more global or networked perspective, then the ROI starts to become really clear. And it's just a matter of packaging that in a way that investors can understand. That's fantastic. It's sort of shifting it from government and payers to maybe to consume.
So I just got an ORR range. And I love it. It's like telling me about my sleep. I'm a terrible sleeper.
I'm not my sleep efficiencies course terrible. But I was thinking about AI and eye care and consumers and maybe not going toward the traditional reimbursement pathway. Yeah, that's right. And bypassing that, the traditional sort of what we think about as eye care providers, but going straight to consumer.
What are your thoughts? Yeah. No, I actually really love that model. I mean, I used to not love it because it was just a different market, I think, years ago.
But now consumers are, you know, our patients are consumers and they've got an ORR ring or an Apple watch and they're tracking their sleep patterns and all that. And they want to understand insights about themselves. The whole longevity movement has really come into fruition. You know, patients are taking supplements and wanting to understand their bodies, doing whole body MRI scans and CT scans where they really don't need them.
Catering to that consumer mentality is really important when you're thinking about creating a startup. And in fact, you know, that can essentially crowdsource, you know, all of your data. So I do think it's a brilliant path, but you have to show real tangible value and you also have to make sure that the technology is very easy to use and scalable. And I think those are really key things really and not just for consumer focused technology, but for any technology.
They should be HR integrated because otherwise physicians are going to have a hiccup in their workflow and anything that rose a little monkey wrench in our workflow, we're just not going to do. And so you can have the best technology in the world, but if it doesn't fit in that physician workflow or that consumer workflow, then it will never be used. So that's another really important lesson. Availability, adaptability to different market conditions.
Obviously, HIPAA compliance, EPR compliance are very big topics. Engaging with the FDA early is also important. They can give you some pre-market assessments about, you know, or pre-submission assessments about what, how you should structure your technology or how you should focus it. And that can be really important.
And then also setting up a mechanism for post-market analysis in real world situations is also crucial. This has been super helpful for me. Where your FDA had to be a consumer to love it all. It's really helpful.
I'm sure it's helpful to a lot of the innovators out there who really we depend on on driving this field forward. So thank you for your thoughts there. I'm looking forward to the excitement. Talking about the whole landscape of AI and eye care for optometry, optical and ophthalmology.
The next topic is something I know actually very little about. It's the metaverse. And this is an equity, we're really delighted. I'm an expert in the field.
Dr. Bosch on the metaverse and ophthalmology in the AI. So that's a really fascinating intersection. Eye care, the metaverse in AI.
So take our hands and walk us through like we're five. Like this is a chat, keep the key problem. What is the metaverse and what's the relationship between eye care, the metaverse and AI? What's AI doing?
Okay, great. Okay, so let me start with calling it spatial computing instead of just metaverse. And I'll tell you the difference. I'll tell you the metaverse is imagine you are within a video game.
You are inside the video game playing. And so everything around you is constantly being generated. So if you turn your head to the side, you're going to see another aspect of the video game, for instance, and it's constantly generating for you. So it's not already there.
It's always being generated. That's where AI comes in. Depending on your vantage point and your perspective is what you're going to see. And that is the same for spatial audio as well.
Spatial audio means that if you're walking within the metaverse or within this video game of yours, you may hear a car explosion over here or, you know, hear a child talking with you over here. But if you move closer to that child, then the voice is going to amplify as well. And so if you think of medical meetings, for instance, in the metaverse, this is a way where you can come closer to me to hear a conversation or back off or go into a separate area to discuss with one of our industry partners, a certain technology. So that's where spatial audio and the metaverse come in.
Now augmented reality means that you're still in your everyday environment, but there's stuff superimposed within that everyday environment. So a really common example of that is Pokémon, for instance. If you ever played that, you would see the little character, but it's just in front of the CDS or whatever it is in front of you or in your living room. And those are the two sort of different types.
Spatial computing takes into account all of those things together so that you can have an immersive, a totally immersive experience if you want or just somewhat immersive experience with augmented reality. And now you're getting a little bit of a feel for what that is and what AI is used for. AI is constantly generating those patterns that are around you. And so as you're moving through real time, the AI is generating the environment around you, constructing the environment continually.
I think this is a relative. Tell me if I'm way, way off here. The meta has come out with their obviously their Ray band, and their work out some kind of, those are all, I mean, those are all sort of I wear devices. Right.
And soon we have a role. I mean, they're kind of on our real estate here. Yeah. I feel like I care providers should have a real big role here.
Yeah. I think these devices alone and think about our sort of optical friends. How should we think about that? Where is that?
What's the role that they're going to be playing? What are these devices going to? Are you talking about wearing a headset at all times? Or do you imagine much smaller form factors in the future?
Walk us through that. So it depends what you're trying to do. If you want that totally immersive experience, you probably need to be in a headset. And maybe I'll give an, I'll give an example here that's really potent.
And then we can work back from there if you'd like. So kids with amblyopia. For instance, what did we used to do when you and I were training or earlier in the past, what we were doing was patching one eye for the most part, right? There's after pain and there's stuff like that.
But most of the time was patching. Kids hate it. And even more importantly, it's harder to enforce, but it is also detrimental to the child because then they never fully develop binocularity, right? It definitely takes a toll on their binocularity over time.
And so with virtual reality based tools, now you can put a child in a headset, have them watch a Disney movie for an hour and a half or two hours. And during that cycle, the stronger eye will be fogged. And that's where the lenses and the opticals come in, right? Because now you have this dynamic lens inside there that can adjust to different lighting conditions and different eye conditions.
And so it's fogging the stronger eye, forcing the weaker eye to then work, but also giving it certain tasks like following a cartoon character across the screen where theoretically, it's more like using a peripheral vision rather than fixating at a certain point. And so those kids are actually developing, they're already showing much better amblyopia success and much better binocularity. It's preserving their binocularity because you're not taking one eye completely out of the game there. I think I know what my kids would probably want.
Yeah, exactly. Or playing a video game in VR or just like I said, just a Disney movie. One of the companies, again, no financial interest here, but one of the companies named Lumenopia just did a, this is a couple of years back, but they did a partnership with Disney. And so it was just Disney movies that were being shown in there, but they were being actively fogged and those children were showing some really great benefits from it.
So that's one thing that I think is really compelling about an immersive environment, just for patient therapies. Imagine a stroke patient. I've had patients before where I've died. I've actually prescribed, almost prescribed a VR headset because those patients have had a stroke.
They're immobile, not interacting with their families, pulling away, had an ALS patient, a stroke patient. And now they're in that headset and they're skiing on the beach and they're skiing in the Alps or something, you know, or dancing on the beach in Ipanema and really starting to enjoy life instead. And so that kind of power in a virtual reality or augmented reality is really important to me. I think personally, I think that's a really important use case for physicians.
That's a really powerful example to think about how it can help patients. I could also think about helping surgeons and I can provide a patient. Yeah. This is where I spend a lot of time, actually.
So now imagine, you know, training a student and they get a 3D stereoscopic view of the surgical field. And with AI, because you were asking how AI relates to this, you know, imagine they're making their capsule rexis and they get a little rent, for instance, right? But this is all simulated environments. So the fluid would rush up or the vitreous would come out, you know, and these are real-time simulations that are now reacting to what that student has already done.
And that's the real power in surgical training or even navigation, which we can talk about in a minute. Navigation is just where you have an augmented reality overlay on top of the surgical field so that it takes the CT scans and MRIs from the patient or any of the diagnostic imaging from the patient and then creates a 3D model so that as you're operating, you can see those different landmarks or pathways and find better ways through them so that you don't hit a nerve bundle or a bundle of blood vessels or whatnot and cause a bleed. Wow. So from patient education to physician education to special computing and conferences, there's probably no connection between, you know, special computing and AI, but clearly there's a huge interaction between all of these three.
But Yuki cannot have special computing without AI. The crux of this is that that environment has to be constantly simulated. It's not already there. So here at Dr.
Bauch, we're going to close out by talking about bringing out the crystal ball and asking you sort of money down on the table and tell us, you know, where you see things are going to go in the next, you know, maybe one year and then AI. If there's, you know, maybe one or two sort of predictions that you want to make for AI and I care, let's hear it straight from you. So what excites me most is the growth of precision medicine. That's really what I spend a lot of time on right now.
And that includes digital twinning, which means that you get a digital representation of you with all the real world data surrounding you, which is constantly informing the system. And that's important because that digital twin takes into account everything that makes you who you are. And that's your lifestyle, your socioeconomic factors, your genetics and also your clinical data, right? And so now you've got this comprehensive digital image of you or digital representation of you, but you can run it through real world scenarios with AI algorithms under different conditions.
If we take that back to an eye care perspective, now imagine you've got a glaucoma patient and you have a digital twin of your glaucoma patient, Mr. Smith or Mr. Jones. And you can put him through different conditions based on his clinical data, his socioeconomic data and his lifestyle data, and then also his genetic data too for glaucoma and or just for really genetics in general.
And now, pry under conditions of one eye drop for glaucoma treatment versus two eye drops versus a valve or a shunt or, you know, it just depends on one year or five years or ten years or whatever, you know, however you want to do it. But basically try it under different treatment methodologies and roll that timeline out based on all of these different algorithms that you can put him through. And then go back and pick your best of your favorite outcome for him and treat him as such. I didn't do a very good job, I think, of describing that.
But basically you get the idea, it's sort of, I call it going back to the future because you're taking these AI algorithms using all of that data that makes that digital twin who it is, but then putting it under different conditions with those AI algorithms, picking your best outcome based on those results and going back and treating the patient accordingly. I think that is super powerful and I really feel like that is where we're headed. I mean, we're right on the cusp of that right now. In fact, it's funny because we've been doing that for manufacturing and supply chain, financial market strategy for a very long time, like predicting stock trends, for instance.
But we haven't quite nailed it yet in medicine. And I think that's really where things are headed in the next, you know, within the next easily within the next five years. Thank you so much. You're always touched a pleasure to talk to.
So thank you so much. You too. And this was a lot of fun. I really enjoyed talking with you.
And you know, we did cover a range of topics and you were just amazing. Your questions are amazing. And the way you think through things, I know that we are on the same page. So I really appreciate you letting me come on to your inaugural podcast.