Scroll Top

Carlos Ciller – AI Is the Window to the Soul at RetinAI

On this episode of The Harry Glorikian Show, I’m hosting Carlos Ciller, CEO of RetinAI. Their mission is to help eye doctors and other professionals in that field get more and better data from new kinds of eye imaging.

Keep on reading through for the full transcript of Episode 103 with Carlos Ciller.

Harry Glorikian: Hello. I’m Harry Glorikian, and this is The Harry Glorikian Show, where we explore how technology is changing everything we know about healthcare.  

And one of the ways technology is changing healthcare is through the explosion of digital images of almost every part of the body. 

There are the familiar types of imaging everyone knows, like CT scans, MRIs, ultrasound, and of course, X-rays. 

But these days doctors and medical researchers are also exploring newer types of digital imaging technology, such as Optical Coherence Tomography, or OCT. 

OCT uses near-infrared light that penetrates just a couple of millimeters into a tissue such as an artery wall or the retina of the eye.  

By collecting the light that scatters back, OCT can produce an incredibly high-resolution cross section or even a 3D reconstruction of the tissue. 

Ophthalmology is one of the fields putting OCT to use most aggressively, partly because it’s perfect for showing cross-sections of the retina, the iris, the cornea, or the lens on the scale of micrometers. 

But as you can imagine, every time an ophthalmologist or optometrist uses an OCT scanner, the procedure generates a huge amount of digital data. 

And my guest this week, Carlos Ciller, started a company called RetinAI whose mission is to help eye doctors, eye surgeons, and scientists studying the eye manage and analyze all that information. 

And not just information from OCT, but from other types of eye imaging like fundus photography and fluorescent angiography. 

At one level, the company is just doing its part to cure a huge headache we’ve talked about again and again on the show, which is the lack of standards and interoperability in the healthcare IT world. 

They want to make it possible to store and analyze digital images of the eye no matter what technology or device was used to capture it. 

But more intriguingly, once that data is stored in a structured way, it’s possible to use machine learning and other forms of artificial intelligence to sort through image data and identify pathologies or double-check the judgments of human physicians. 

RetinAI is developing algorithms that could make it easier to diagnose and treat common conditions like age-related macular degeneration—a form of damage to the retina that causes vision loss in almost 200 million people around the world. 

Ciller told me he started out his career as a telecom engineer and never thought he’d wind up running a 40-person company that works to help people with vision problems. 

But at a time when there’s so much new data available to diagnose disease  rand identify the best treatments, journey’s like Ciller’s—from the computer lab to the clinic—are becoming more and more common.    

Here’s our full interview. 

Carlos Ciller on the Harry Glorikian Show

Harry Glorikian: Carlos, welcome to the show. 

Carlos Ciller: Thank you for having me. 

Harry Glorikian: So. I was really excited about this when I when I found the company and what you guys are doing because I’ve always been, you know, fascinated by ophthalmology and so forth. But and I’d like to, you know, get to the question about basic unmet need that, you know, your company, RetinAI is is meeting in the world of ophthalmology.

But before we even go there, because we’ve got a lot of different types of listeners on the that are all over the world that are listening.

What is ophthalmology?

What are the parts of the body, what are the diseases that, say, an ophthalmologist is looking to treat? I mean, it’s everything to do with the health of the eyes, from cataracts to glaucoma to retinopathy. But maybe you can give people a brief overview. 

Carlos Ciller: So I think I will try my best. So ophthalmology is the science of the eye. And I think it’s a beautiful science because among all the different senses, it could be one of the most important. And if you think about it, you you always are going to have, I mean, you sometimes it takes some time to value how much something is good until you lose it.

So losing your eyesight is one of the worst things that can happen to you. So it’s a very important thing that often you don’t recall or you don’t remember. And a bit more about ophthalmology. So we have to go back millions of years and there was at the beginning of the Cambrian evolution.

So 550 million years ago where the first sensors were actually developing in order to capture light. And that was actually happening in very small organisms. And then over millions of years it actually evolved to cover those sensors because they were very sensitive.

And then it has evolved into, into the marvel of science that we have today with human eyes, with bird eyes, which had very advanced species of engineering. And it’s actually a very precious organ, so precious that we have actually devoted myself personally and the company as well, a very big part of our lives to it study. 

Carlos Ciller: So and we so for instance, a bit about my background. So I basically did a PhD in machine learning, applied to medical imaging ophthalmology. That happened close to ten years ago. So I started back when AI was not as hot as it came afterwards in 2014.

So it was and it just came by accident that basically I maybe there will be some questions linked to that, so I won’t go too much into into detail. But, but it has been a very nice encounter.

And when it comes to diseases, most of your audience may know about the typical. So cataract surgery is one of the most common eye surgeries that is other different types of. So you have front of the eye and back of the eye. Normally ophthalmology is divided into these two big areas.

It’s kind of the two sides of two different teams in in a football match that you may think of and especially for front of the eye, you have cataract disease, glaucoma in some cases, which is one of the diseases that physicians and researchers know the least. So still, there is a lot of ground to cover when it comes to glaucoma treatment, origins and how the disease evolves. 

Carlos Ciller: And then there is, of course, the back of the eye. And retina is a very a very wide field where you have vascular diseases or diseases that are affecting the vascularity of the eyes. Then you have the age related macular degeneration with two different forms. So you have wet age related macular degeneration, for which there are many treatments in the market today by some of the pharmaceutical companies.

Dry AMD, which is an area of unmet need where today there is no there was no treatment up until quite recently, where FDA is now approving some of these new therapies that are that are coming out. And it’s a very exciting moment to be specifically in dry age related macular degeneration.

Then there are other diseases, diabetic retinopathy, diabetic macular edema, retinal vein occlusion, and a lot of genetic and rare disorders. So you have the eye is also an area where you have tens actually maybe close to 50 different type of genetic disorders that have been identified.

So there are a lot of new upcoming gene therapies that are developed, that are being developed and that hopefully will be able to take some of these very rare disorders or target disease, retinitis pigmentosa, just to name a few. 

Harry Glorikian: Yeah, we take we take a lot for granted with this this optical thing that, you know, there’s so many things that can go wrong. But so but in reality and I’ve got a few friends that are in the field and we always talk about these things.

I don’t think people realize how high tech a lot of ophthalmology clinics have become over time. I. I mean, maybe you can talk about, I don’t know, some of the different technologies or and data that are commonly used in the field today. 

Carlos Ciller: So you have, it has been, ophthalmology goes back to the early forties. And I won’t give the whole the whole one hour description of the evolution of the field. But you have different forms of imaging that are normally being used to make different types of diagnosis.

So today you have very low cost machines or images that are being captured from the back of the eye. For instance, fundus image photography, fluorescent angiography that where you inject that eye, that is specifically going through the vascularization of the of the retina.

And then you have more advanced pieces of technology such as optical coherence tomography. So OCT, for those of you who are in the medical domain and haven’t heard about it before, it’s something very similar to an MRI scanner, but for a very small region of the eye.

So you could imagine like a very small window of one to maybe three to up to 12 millimeters today with a very high micrometer resolution with the different layers of the retina, it looks like a bit like a lasagna as we say. And then you can basically be able to navigate. So you have OCT in different forms, sorry, OCT and different forms of capturing OCT.

And you also have now more recently OCT, which is looking at the vascularization of the retina. So you have a very wide spectrum that goes from low cost $500 to $1,000 USD medical devices all the way to OCT machines or OCT machines, where certain pieces of technology that can go as high as $100,000 to $150,000. 

Harry Glorikian: So you know you’ve mentioned you know OCT. Right. And you’ve mentioned fundus photography right. I mean this is where I’m sort of going to draw a line to where, where you guys are. But there’s a lot of images there, right. And I know a little bit about the imaging business, right?

I bet those files are all in a different formats depending on the manufacturer. We talk a lot on this show about how medical fields where data management is a challenge, but is this why ophthalmology has become one of those fields? So is that the unmet need that you guys are trying to to solve by one of these unmet needs?

In other words, are there data related obstacles holding back better diagnostics or better outcomes for, you know, patients? 

Carlos Ciller: So one of the I think one of the most important pillars and we as an organization we also hold for those pillars is interoperability. And interoperability up until which is being able to bring one where the information that is maybe located in one device into another device or a different platform or even a patient that would be able to take that data out of the device to be able to go to a different platform, to be able to keep on operating with a platform.

Ophthalmology is quite behind in this regard. I would say something in the range of 10 years behind the radiology space when it comes to having these standardization and interoperability standards very well defined.

So ophthalmology, even if there has have been multiple attempts at creating standards to be able to have different imaging solutions in ophthalmology, those have some, haven’t been adopted by medical device manufacturers in the past.

Now with FDA put in a bit more pressure, it’s becoming more and more a real need because the reality is that, yes, you have a lot of backstory, maybe decades of data that is stored in proprietary formats and that is not easily accessible. 

Carlos Ciller: And for us and I can tell you about personal pain, during my Ph.D., I was actually operating with these type of situations.

But in a way, the company has been the response to our own frustration. I think the best things come out of frustration. So, yeah, so doing a PhD in this field, seeing that there was this difficulty to bring innovation faster to to the people who need it the most, basically the patient.

That was a source of frustration that actually brought us, myself and my two co-founders to build the company. And that is one of the main problems that we are solving today. Some people believe that AI is the big problem to solve. And that’s actually not true. AI is just the cherry on top of the cake.

The biggest problem that you need to solve first is the data management problem. Once you solve that, once you are able to harmonize standardized and properly structured data, the AI part is just unfortunately, it’s just a means to an end for me, even if I’ve done a PhD in that field, which is a bit sad. 

Harry Glorikian: Well, you know, it’s I can tell you that on I would say 80% of the shows, this is a a theme regardless of the medical area. And I just it baffles me that we can’t sort of standardize on certain things and and the government don’t seem to understand that they need to apply the pressure to get people to standardize on this.

But I want to get to how you’re addressing those unmet needs. But first, let’s go to, I’d love to talk a little bit about the RetinAI origin story. Right. You’ve mentioned you were a PhD candidate at the Center for Biomedical Imaging at Lausanne University. And you you were working on this computer assisted treatment planning system for intraocular tumors. And you’re. Is that where you met your cofounders, Sandra De Zanets and Stefanos? I’m not going to try to pronounce his last name. 

Carlos Ciller: I can I can do it for you for for, for posterity. So it’s Sandro de Zanets and Stefanos Apostolopolous. So I can, I can give a bit about the backstory. Originally I was not planning to go towards the I, I was actually I started, so I’m an engineer by training. I did telecommunication engineering in Spain where I, where I’m from. And then a bit more than 10 years ago, I came to Switzerland because I wanted to see the world.

And I finished my my master studies at Ecole Polytechnique Federale Lausanne in Switzerland. Then I started, I was supposed to start a PhD in a specific area, but then this new opportunity to specifically work in patients with retinoblastoma tumors, so basically in children with retinoblastoma appeared.

And at the beginning I was not very sure that this was going to be interesting for me, but I gave it a try and I saw the opportunity because the PhD was across multiple Swiss institutions.

Retinoblastoma is not one of the most common tumors or in most common cancers. It’s actually the prevalence is one out of every 20,000 children are born with retinoblastoma and it’s a devastating disease. It’s really bad for for everyone. The socioeconomic impact is huge.

And when I started living, during my PhD, defining new methods to improve the treatment therapies for these children, I realized that it was actually a very meaningful way of spending your time. 

Carlos Ciller: So if you can contribute to make the life of other people better, that’s a very good reward. And and then the first year I discarded the idea of becoming a professor because the academia world is so difficult that then I say, I think I’m going to I’m going to have a very hard time.

And then at the same time, I met Sandra and Stefano, who were fellow students in the University of Bern in the ARTORG Center from the University of Bern. And we share very common. We were always speaking about different areas. We are really much alike. We come from very different backgrounds or very different geographies.

So Stefanos is Greek, Sandro is Swiss-Italian, and myself, I’m Spanish, but we got to understand each other very well. And back in 2014, we there was the opportunity to go together to a hackathon. So there was I don’t know if you’re familiar with hackathons or it’s basically a weekend where you spend all the all the weekend coding and then and then you. So we went together.

We spent 48 hours coding with only 2 hours of sleep, and we worked on an application. It was a social network back in the day when social networks were still, I think, geo localized, anonymous message board. 

Carlos Ciller: And what we tested was that we could work together on whatever the project. And even if the idea that the idea didn’t take off, we saw that we could work together and then we just needed something to work on and that something to work on was actually the feel that we were all doing our PhDs on.

So we wanted to make sure that innovation that was being developing and developing in the lab was going to make it to the people who need it the most.

And sometimes it’s a bit personal, that we had this naive bias. We didn’t really understand how difficult that was going to be. So we actually put all our savings into the company. This is not a joke. It’s it’s fun. And then we went for it. And then the idea was clear was to make sure that the lives of patients were going to be better by breaking the technological barriers.

And we were three at the beginning, but now we are close to 40 people in the company, and I’m very excited that we are growing every day and breaking more and more barriers to make them to make the life of people better. It’s very rewarding. 

Harry Glorikian: So if every entrepreneur understood exactly how hard it was, they wouldn’t do it. 

Carlos Ciller: Nobody. There would be no startups. I mean, yes, yes. 

Harry Glorikian: Yes, yes. And there’s so many companies I’ve started that, you know, I’m like, Yeah, okay, that’s doable. And then you stand in, you’re standing in the fire and you’re like, How the hell did I get here? But once you’re in it, you have to you have to sort of break through to the other side.

But if you knew how difficult it was. You know, I’ve started a few businesses where I’m like, Nah, I wouldn’t do that again, I wouldn’t do that a second time. But. Okay, now, let’s, I would love to start talking about the actual products. I mean, the main one is called Discovery, if I’ve gotten that correctly.

And as I understand it, it’s a platform for aggregating and harmonizing ophthalmology data from lots of different sources. But what does it actually do? What kind of problems is it designed to solve in the lab? And do you have different versions? You know, is there one for the patient side? Is there one for a clinical researcher? Is there one for drug developers or is it all just one big platform? 

Carlos Ciller: Yes. So Discovery is the is the accumulation of all these [inaudible] from us. And it’s actually our response to the problem of data management and the common denominator across all the different stakeholders in healthcare. Be it pharmaceutical companies, be it hospitals, be it even patients.

In the future, as we move more towards a patient-centric data control is data organization. So that was actually the first problem to solve, and that is what Discovery is solving. Being able to collect data that is being generated as part of with an imaging device, electronic health record data, demographics, even genetic data, all of these can be collected, funneled into Discovery and structure for on a per patient basis.

And once you have done that, once the data is properly organized, we have a different ledger where we come in with our algorithms in order to enrich that data. And then here it’s the first categorization. So we have today, Discovery the platform is certified as a medical device for both the European market, and we receive FDA clearance back in beginning of May this year.

So we are very excited about that, that we can also commercialize our products in the US. And some of the AI algorithms, they come on top of this data organization and then you extract additional insights. And different stakeholders have different interests in some of these insights. 

Carlos Ciller: So for instance, for pharma, Discovery is being used today for two main purposes. On the one side, pharmaceutical companies in the public domain, like Novartis, with whom we have a relationship.

They are using it to support their own internal research and development pipeline, making decisions on a daily basis of what is what are the different patient populations that that we are looking at for the next compounds, which are the different clinical studies that we need to organize our different compounds behaving.

And of course you need to understand that they have tons and tons of data from multiple clinical studies. So you can just put all this data into discovery and have this collection of information that remains. So even if the people change and different teams are using it, so that remains.  

Carlos Ciller: An extension of that version is discovery for clinical studies. So while Discovery Unity, the first one for internal research and development is used by pharma, clinical studies enables you to run decentralized clinical studies at scale. And again, the same problem remains data organization, analysis, data collection.

And they we have a different type of flavor that enables you to do real time management and monitoring of the clinical study. So here we have two products that are for a very specific segment. And of course, if you remember our ulterior goal, we want to make sure that we elevate the quality of care for patients. 

Carlos Ciller: So there is no reason to stop there with our products. We have now in the last over the last couple of weeks, we are releasing our first clinical product or for the clinics and hospital segment. It’s called Discovery Core. And it’s a clinical academic research tool that can be used to basically support your internal research.

So people work in ophthalmology, physicians that are very busy in the in the daily routine that they want to start doing some research, but they don’t have the time. Hopefully Discovery Core is going to make it very easy to work with peers, to some to use some of our algorithms on top of that, and basically increasing the pace of innovation of the field as a whole.

And then you can imagine that because they are relying on some similar background technology, then we can just connect the dots. And that has a lot of potential additional value generated.

So collecting real world evidence, being able to accelerate the development of clinical studies by being able to very quickly identify the patients that could be candidates for a clinical study and accelerating the whole process that five years ago we didn’t know we could do all of this, but this is the direction that we are going to.

And again, common denominator data management after as a main problem to solve. 

Harry Glorikian: Let’s pause the conversation for a minute to talk about one small but important thing you can do, to help keep the podcast going. And that’s leave a rating and a review for the show on Apple Podcasts. 

All you have to do is open the Apple Podcasts app on your smartphone, search for The Harry Glorikian Show, and scroll down to the Ratings & Reviews section. Tap the stars to rate the show, and then tap the link that says Write a Review to leave your comments.  

It’ll only take a minute, but you’ll be doing a lot to help other listeners discover the show. 

And one more thing. If you like the interviews we do here on the show I know you’ll like my new book, The Future You: How Artificial Intelligence Can Help You Get Healthier, Stress Less, and Live Longer.  

It’s a friendly and accessible tour of all the ways today’s information technologies are helping us diagnose diseases faster, treat them more precisely, and create personalized diet and exercise programs to prevent them in the first place. 

The book is now available in print and ebook formats. Just go to Amazon or Barnes & Noble and search for The Future You by Harry Glorikian. 

And now, back to the show. 

[musical interlude] 

Harry Glorikian: So, you know, my questions are like, you know, what kind of machine learning or AI techniques are you using? Right? Where does the training data come from? Because you’ve got to get these things spun up. And if you could give maybe a couple of examples of what kind of insights do the algorithms generate? 

Carlos Ciller: So for instance, I would characterize at least three different types of insights. So you have the basic insights of classification. For instance, you have a patient that maybe comes in. Let’s say there are multiple FDA approved algorithms in this regard for detection of more than mild diabetic retinopathy.

This is a classification task. You basically say diabetic retinopathy. Yes/no? And that is actually taking an image as an input. You have different algorithms that are analyzing the different patterns within the image, and they have been trained in in a way as to identify those patterns. And so that is. That is done. That is a specific type of machine learning model. It’s a classification model. 

Carlos Ciller: Then you have segmentation models, segmentation models are going to be a bit of drawing this overlay of a specific area. So for instance, you could have segmentation of the amount of fluid within that retinal image. So if you have enough image for a diabetic retinopathy patient, the amount of fluid for the micro aneurysm.

You will be able to potentially quantify the amount. That of data specific, a pathological biomarker. And we have some of these algorithms for both two dimensional images as well as three dimensional volumes. So one of the things that is specifically unique that we have certified recently is the possibility to quantify the amount of fluid in different retinal layers. 

Carlos Ciller: So for instance, if you have a drug or neovascularization and you are able, one of the potential things that you’re looking at is decreasing the amount of fluid in the retina. You are able to quantify this decrease as a result of the treatment, which is something that before you couldn’t do.

So we have classification, you have a segmentation, and we have some models that are specifically linked to progression prediction. So let’s say that you have a patient coming into the clinic for geographic atrophy, just to give an example.

And with our models, you can identify whether a patient is going to be a fast progression. So basically the degeneration is going to move faster on this patient versus a slow progression.

And if you look at the other side from the payers’ perspective. When they have to make decisions on who is going to get the new drugs that are coming out into the market earlier these fast progresses who are going to benefit more.

But from the drug are potentially targets. So having progression algorithms gives you an advantage to identify those. And then of course, we have not so sexy algorithms as well working in the background to identify image quality. 

Carlos Ciller: So let’s say that somebody is collecting, you have somebody else, not the physician collecting the image, and then you want to make sure that the image quality is good enough as to be assessed.

So we have algorithms doing that. And this is super important because for clinical studies, you don’t want to make the patient come for a second time because the image is unreadable. Just to give an example. So and then we have other algorithms, but these are more in the background to measure the uncertainty. So for instance, how uncertain or how certain that you are about your segmentation.

For instance, let’s say that you have an algorithm that has a very difficult case of a patient with very low quality, and the algorithm is not certain of the segmentation that is making of the delineation that it’s making. You are able to measure that, and that can be an additional metric. And we have some algorithms in that direction, too.

But yes, so you use, again, I think, a today until we reach artificial general intelligence, and I think we are still far from it. It’s more the means to an end. We identify a problem and then we just look at the collection of algorithms. How you can get that. 

Harry Glorikian: So now maybe there’s an overlap here. And, you know, I was trying to absorb as much as I could from what was available. But you guys have the software you’ve built for the labs, but then you’ve created, I thought, some specific apps for the clinics that can help doctors catch problems.

Maybe some of the stuff you were talking about was overlapping with that. But, you know, is there specific types of eye diseases that you’re looking for, the algorithm can detect. I mean, like you said, you have this model that can measure thickness of the retina and layers of the eye. And, you know, that can flag excess fluid accumulating in those layers.

You know, one of the things we didn’t talk about was, you know, how did you get the training data for this? And yeah, are there any other diseases? We haven’t sort of touched on that these apps for a clinician can find. 

Carlos Ciller: Yes. So maybe the normally the way we train our algorithms is through collaborations with research partners. So we have had over the years since the beginning of the company, we believe that having a collaborative approach, working with research institutions, just of course preserving the integrity of the patients and the anonymity of the patients, being able to work with these databases of data in the context of research and afterwards being able to to bring it into into a finalized solution.

So today, most of our data sources are coming from these collaborative, collaborations that we have with different research partners. Then when it comes to, and we are very happy because in a way you can contribute to science because we publish everything that we do, you are able to support physicians because they can just basically keep on developing science and they see the return.

The patient is going to eventually have the return as well when you have new therapies that are being supported. So we think that this is a model that works very well for us when it comes to the specifics of diseases. The way we look at the problem is by creating a library of biomarkers. So for instance, let’s call it, let’s say vascular and vascular AMD is a specific condition and there are different biomarkers that are manifesting. 

Carlos Ciller: So you could imagine that it’s a supermarket. Maybe this is not the best analogy that it’s like you have like a store and you have different conditions that you are able to identify. And then the collection of these conditions, there’s a grouping of these conditions is going to facilitate the diagnosis of new vascular entry.

So we are creating a library of AI based biomarkers. For instance, retinal layers is one of them, and you have many different retinal layers and more to come, even with better quality devices or more granularity, different conditions such as there are, for instance, geographic atrophy, a drusen, which is like special mountains that are forming in the retina. You combine all of these different biomarkers and a collection of those or a subset of those.

We put together to support the diagnosis or the management of a specific disease. So I can give you an example. For vascular AMD, you it’s very interesting to look at the intra-retinal fluid and subretinal fluid because those are different conditions that are going to manifest during the treatment of the disease or whether you are reacting or whether actually the condition is evolving and then you’re going to use the readouts of these conditions to make decisions on the treatment. 

Carlos Ciller: So what we, the beautiful thing is that we started in ophthalmology, but this library has applications beyond ophthalmology into neurodegenerative disorders, for instance, looking at different retinal cardiovascular conditions.

So for instance, if you have a specific vascularization of your retina, it’s going to be affected in case you have different vascular conditions, have conditions and machine learning can help a lot to be able to use the library of biomarkers that we have. To be able to identify even doing early screening of patients or even monitoring of patients who are already whose condition has already been identified.

So a lot of opportunities and specifically so if we were to name some of some of the conditions that we are treating today in vascular, the or that we are treating that we are supporting in vascular AMD, AMD as a whole, wet-dry AMD, the diabetic macular edema, retinal vein occlusion, diabetic retinopathy, and many more. Glaucoma is coming next.

There are many conditions and little by little covering the entire ophthalmology space. And of course, we won’t stop there. 

Harry Glorikian: Yeah, well, they say the eye is the window to the soul. Right. So we can figure, we can see a lot of things that are happening.

But like, you know, we’ve done quite a few shows here. And so in the past we’ve been talking about, let’s say the digitization of radiology and pathology, because that comes up the most and that creates a big opening for computer vision based analysis of medical images.

And the idea continuously comes up that A.I. can help reduce false negatives by making sure human radiologists or pathologists don’t miss anything important, like a cancer cell. I always tell people, the machine didn’t play poker last night, didn’t have an argument with its spouse. Right.

So I’m wondering if the same principles apply in your area. Are there conditions that, say, human clinic clinicians might miss that the algorithm catches? 

Carlos Ciller: Absolutely. I fully agree with you. One of the things about the algorithms is that you can ask the algorithm again the next day. And at least without algorithms, it’s going to give you the same answer. So that is not going to change. And these, it is not always the case with humans because there are so many additional factors that play a role. And I would even include a different dimension.

There are very well known biomarkers where there is consensus. And the different physicians agree on the specific biomarker. So if you look. If you choose 100 physicians in ophthalmologists, retina specialists in Europe and the United States, and then you ask them, is this intra-retinal fluid? Yes. No.

There is a high chance that they, many of them, agree on what is intra-retinal and what is not. Now, there are other conditions that is a bit of a gray area. So what is intermediate region of atrophy. What is the, what is an outer retinal atrophy. Which is the boundary of the end of the atrophy and the beginning of the healthy tissue.

You are going to have for every physician a different answer. And there is actually a different. I believe that one of the beautiful things about algorithms is that and I see that you need to start walking before running. I think that now more and more physicians understand that AI algorithms are not here to replace them, because a physician’s job is a combination of tasks. Each of these tasks. In some cases, they may evolve in the future.

So they may be automated or even they could be supported. So you could imagine that, or at least the way I see it, with algorithms going into the clinic, that first it’s going to help you agree on a standardized assessment metrics. And you could have like little helpers that are going to help you double check and make sure that you have your different ideas right when it comes to assessing a patient.

So I think they are going to help us standardizing treatment, making decisions faster and in some cases automating part of the work. Because if you have to look, and I think, and I’m sure that in this show you have seen many other people working in the radiology space, you have to look at the whole history of this patient.

With so many visits with a tumor that is growing or variety in size over time, being able to have this clarity on the small menial tasks. That’s going to be very helpful. And it’s going to, of course, support your decisions. And what we have seen, at least in our collaborations with physicians, is that they are very happy at the beginning.

They are a bit skeptical, but then they are very happy to have the support and leave the menial tasks for the algorithms. 

Harry Glorikian: So that was going to be my next question. So how does what you’re providing change the ophthalmologist’s workflow during or maybe even after the exam? 

Carlos Ciller: So today we don’t change it much because you have, we want to, we took this approach that having a frictionless integration is better than adding additional software into the clinic. It doesn’t really work for us to bring new software into the clinic and try to basically make the work of an ophthalmologist more difficult.

So we are partnering with already existing players in the clinic, medical device manufacturers, companies in order to have frictionless integration and to leverage each other’s strength. So for instance, in the public domain, we have a relationship with Heidelberg Engineering, which is one of the top companies in medical device manufacturers of OCT machines.

Today, a physician from the clinic with no changes can update the latest version of the software, get a patient [image] that can drop it to our logo in the little place. The image is anonymized and de-identified at that point, sent to our servers for processing, and they receive a report back in a few seconds and they don’t have to even change machines. And this is actually the way we believe it’s going to be.

Trying to change completely the workflows is going to be very difficult. By having an integrated workflow where some of the tasks, reading tasks have already be done by somebody else, and you just look at the report.

That has the potential of improving and making the life of many people easier, especially with a new treatment. So if we if we go back to geographic atrophy or a more dry age related macular degeneration and geographic atrophy being one of the conditions of that, there is going to be a lot of education going around the physicians. There is no treatment today. So they will also need to learn with the appearance of these new drugs.

We have an opportunity to help them through this. And I think that that is going to be also an opportunity to increase adoption and improve, of course, outcomes for patients. With the support of a network that is helping. Yeah. 

Harry Glorikian: So that drives me to the, you know, next question, which is, you guys also have an effort to use the technology to support precision medical treatment diseases of the eye. So, I mean, I guess my first question is, do we have enough different drugs for eye conditions that knowing more about the patient or their genotype or their biomarker would make a difference?

And what kind of treatment decisions does the RetinAI Discovery platform help with? 

Carlos Ciller: So one of the things that we are doing is we are partnering with pharma companies and life sciences companies in order to elevate the quality of the drug that is going through the door. And that’s normally what happens is that you have the performance of a drug during a clinical study and you have this level of quality.

Then it goes out the door and then there is sometimes a decrease in the performance because physicians don’t follow exactly the same treatment regime that is in the label or and then and then you see what’s going on here. Why do I get basically these amazing results in the clinical study and then we go out in the real world. And then there is a decay. 

Harry Glorikian: Yes. Exactly. 

Carlos Ciller: So normally what happens is that in some cases, there is some missing information that you may not be cognizant about that could be specific to the clinical study. And we believe that every patient is different. But there is only so much you can know from as part of a clinical study.

So we are working with some part of my customers in developing joint solutions, where we combine all the knowledge that we have with our algorithms and progression prediction algorithms together with a knowledge of how our direct is working to create digital precision medicine solutions, which is basically a digital solution that could be distributed at the point of care.

Elevating the quality of the health care for that specific patient, because the physician will be able to, again, through the same process that I can drop a patient. And be able to know what is the best way forward for that patient. 

Carlos Ciller: Given all the knowledge from the clinical study, of course it’s going to be eventually the decision of the physician. But if you can encode all the knowledge of the clinical study, put it in a digital tool and distribute it at the point of care, especially when you need to bridge that gap on education on how to best treat a dry AMD patient, you have an opportunity to elevate the quality of care from the get go.

And you can imagine that a lot of pharma companies and a lot of different health care stakeholders are very interested in having this because everybody wins. Pharmas are going to produce better. Physicians are going to be able to better use the track or identify when the patient is not reacting to the drug, especially if there are different alternatives, being able to switch treatment earlier and know when to switch treatment.

Payers are going to be happier. And eventually the most important part, patients are going to be happier because they are better treated. So we see that, that to me is one of the goals that we didn’t know initially when we started the company. We wanted to break technological barriers, but we want to make sure that we get that, because this is the future of precision medicine.

You as a patient, you have your own treatment that is unique to you, that is validated, that is verified, that is regulated. And there cannot be any mistake. And I think that is a that is a very nice milestone to achieve. We we are hopefully not so far. 

Harry Glorikian: So I was also seeing on the company website, you talk a little bit about how I could help promote the use of telemedicine in ophthalmology, help more patients get access to eye care, especially during periods like the COVID pandemic when people can’t always come to the clinic. Can you talk a little bit about how your software can make tele ophthalmology appointments more productive? 

Carlos Ciller: Yeah. So actually the good thing is that through automation, we believe that there is going to be a disruption in the health care workflow. So today you’ll visit a lot more an optician or a pharmacy than do you visit a cardiologist. Normally. Normally that is the case.

So we believe that teletechnology, being able to go to your local optician or to your local pharmacy, being able to have a closer follow up and monitoring of your chronic conditions— because most of the conditions in ophthalmology are normally chronic, so there is no real cure—there is an opportunity to alleviate the growing number of patients visiting ophthalmology clinics.

So you could just make sure that the patients that are visiting you are those that are actually in need of having a treatment at that point in time. And you can alleviate that by having by having this type of tele-ophthalmology assessment where you could go to as a user to annotation. And this is happening already.

You get an eye scan. A solution such as our platform could be just collecting the eye scan. And because we are also present in the clinic, we can make sure that there is this health care continuum where this patient is collecting data and this patient can actually bring that information to the physician.

So you don’t have to go with papers or with pictures that you have taken on your smartphone to the physician with incomplete and navigation capabilities on the data from your last visit. And we believe that the good thing is that today the platform is ready to do that.

That’s the reason why we are also certified as a medical device so we can enable that type of use. And even in the future, you as a patient, you could be the one in control. So you could think of something like a mobile phone. You go to your optician and you just put your phone there to collect your retinal scan.

Your physician, who is already connected to the platform, can receive the data that that that you just collected and they can tell you whether you have to go to the to the doctor or not. And all of these needs us to be both in the in the phone or in the in the optician chain or in the pharmacy and also in the clinic.

But we are slowly going to get there. So one step at a time. But I, I am very excited about the opportunities that this will open for patients to. 

Harry Glorikian: Yeah. I mean, it’s funny because I’ve been giving I you know, I’ve written a few books about how data and, you know, technology are changing how we’re going to impact health care, how it’s going to change life science research. And it’s in some of my talks, I talk about how you’re going to you might, you know, very soon be able to walk up to a kiosk or something that would take an optical scan.

And then if it identifies something through a, you know, telemedicine platform, you can interact with a physician and then be put on the right track as opposed to wait in line, get an appointment six months from now, you know, and and deal with it then.

So I do believe that these technologies will have a dramatic impact, especially anything image based where you can really move the needle a little bit faster, because we’re getting better and better at that every day. 

Carlos Ciller: Absolutely. Yeah. I fully agree with I share the same view and of the future. And healthcare is going to come to you. In the past, you were going to the hospital to get health care and you were kind of just a spectator. Just looking at health care and somebody else taking control of your health.

Now it’s a joint discussion where you are also in control as a patient, as a prospective patient. And that is changing. So eventually patients will be in control.

They will be able to have better engagement, better. Better follow ups of their own conditions, and especially when it comes to doing early screening of some of some of the diseases or even diseases that could be monitored or screened through your eyes, such as neurodegenerative disorders or cardiovascular conditions.

The sooner you start treating them, the better is going to be for you. If there is a treatment for for Alzheimer’s one day that is working, wouldn’t you be willing to have that treatment? The sooner that you can. So I think that is. 

Harry Glorikian: Absolutely. 

Carlos Ciller: Exactly. 

Harry Glorikian: I can’t it can’t move fast enough, in my humble opinion, because I’m getting older, so I want the stuff to move as fast as possible. You guys were born in Bern, Switzerland, but in September, you guys are becoming neighbors here in Boston. Yes. I mean, I know why you guys started in you know, why you were going to school there and so forth.

But is it is one of the advantages just being closer to pharmaceutical companies like Novartis or, you know, And then on the flip side, why Boston? You know, is there different functions in different areas or why? Why why the what really makes the two centers of of excellence, let’s say? 

Carlos Ciller: So I would say, first of all, Boston and I don’t know about your audience distribution, but Boston is one of my favorite cities in the US. I think it’s a it’s a very beautiful environment. It’s also very European, and I’m European myself, so I feel a bit more at home and at the same time I think it’s a fantastic life science hub.

It’s only 6 hours difference in terms of time, time shift as compared to our headquarters in Bern. So I think it’s a it’s a very good match and we are very serious about our expansion to to the United States. Of course, there are a lot of pharmaceutical companies, both in the New York and New York area as well as in Boston area. So I think that the East Coast is the place to be for us at the moment.

And at the same time, we we are going to expand our commercial operations in the United States in the years to come. So I think we need to be present just to make sure that we can go they can take a car or maybe a plane, because distances in the United States are quite far from each other. 

Harry Glorikian: A little bit longer. 

Carlos Ciller: Yes. So take a plane and then just go and meet your customers face to face it face to face. Even today with with you can always meet remotely like we are doing today or through a digital platform. But it’s much easier to and it’s part of our expansion plan.

We want to we want to be a global company and being in Boston is one of the next steps. So I’m very excited. And to grow the office in Boston, especially in 2023, we are going to increase the number of people there and we will be neighbors so we can go for coffee more often. 

Harry Glorikian: Okay. I always like to keep on top of all the companies that I talk to and interact with. But, you know, just to wrap it up, like I want to step back for a second and just talk about the average consumer, right. And the world of and what’s going on in your world of data management for ophthalmology.

I mean, if I’m hearing you correctly, do you believe that we’re near a point where it’ll be easier to detect and treat most of the important eye diseases and is is having the better software one of the keys to getting there? 

Carlos Ciller: I think having having the better software is definitely going to help. Having the right partnerships and the right relationships or being present in the right locations is going to help even more than having the right software. Because sometimes you and every health care system is a bit different.

I would say that having the right software is going to bring you there. Having the right partners is certainly going to bring you that faster. That would be my, my, my take. And at the same time, I see that now pharma companies are more and more realizing about the value of going into a digital direction.

The diseases of the future are patient specific, so there is a lot of gene therapies that there is more and more low cost gene sequencing. So I believe that the next 10 to 15 years of pharma development is going to go with very targeted therapies to very targeted populations.

So being connected, being present in the right spot to identify those patients early or making sure that they start going through the pipeline in a very cost efficient way. All of these will help make the life of these people better. That’s and we will work very hard to make sure that that happens. 

Harry Glorikian: Yeah, these are all the subjects I try to cover on on the show in various ways through different angles, but they all come down to the same answer. So a great having you on the show. I mean, I can only wish you luck because like I said, I’m getting older and I, you know, making sure that I can see properly is is a is a key factor.

Right as you as you’re getting older, like if it’s dim light or whatever, I’m like squinting and have to put on my glasses. But I wish you great success and you look forward to getting together for coffee when you’re here. 

Carlos Ciller: Yes, I would make sure we meet. We meet when I’m in Boston. Back again and again. Thank you for having me. It was very nice. 

Harry Glorikian: Thank you. 

Harry Glorikian: That’s it for this week’s episode.  

You can find a full transcript of this episode as well as the full archive of episodes of The Harry Glorikian Show and MoneyBall Medicine at our website.  

Just go to and click on the tab Podcasts. 

I’d like to thank our listeners for boosting The Harry Glorikian Show into the top two and a half percent of global podcasts. 

To make sure you’ll never miss an episode, just open Apple Podcasts or your favorite podcast player and hit follow or subscribe.  

And don’t forget to leave us a rating and review on Apple Podcasts.  

We always love to hear from listeners on Twitter, where you can find me at hglorikian. 

Thanks for listening, stay healthy, and be sure to tune in two weeks from now for our next interview. 

Glossary terms for this episode with carlos ciller


Ophthalmology is a branch of medicine that deals with everything connected to the eye – from anatomy to physiology and diseases. The doctors who are experts of treating eye and visual system conditions are called Ophthalmologists. They can provide a range of services, including prescribing eyeglasses and contact lenses, performing surgical procedures, and prescribing medications to treat eye conditions. 

Ophthalmologists may also work with other healthcare professionals, such as optometrists and opticians, to provide comprehensive care for patients with vision and eye-related problems.

Medical imaging

Medical imaging refers to the creation of images of the inside of the body using different tools and tech gadgets. These images can be used to diagnose and monitor medical conditions, as well as to guide surgical procedures.

Medical imaging is an important tool that supports diagnosis and treatment of many medical conditions. It is often used in conjunction with other diagnostic techniques, such as blood tests and physical exams, to help healthcare providers make an accurate diagnosis and develop a treatment plan.

Vascular disease

Vascular disease is a term used to describe conditions that affect the vascular system or blood vessels in the body. Vascular diseases can occur in any part of the body and can have a wide range of symptoms, depending on the specific condition and the location of the affected blood vessels.