In this episode of Clinical Trials Spotlight, host Andrew Pucker, OD, PhD, FAAO, welcomes Ram Yalamanchili, founder and CEO of Tilda Research, to discuss how AI-powered teammates are streamlining clinical trial operations for sponsors, CROs, and research sites. The conversation explores what the rapid adoption of AI means for the future of ophthalmology and optometry clinical research.
Andrew Pucker, OD, PhD, FAAO:
Welcome to Clinical Trials Spotlight, the podcast where innovation, research, and patient care comes into focus. I’m Andrew Pucker, Chief Development Officer at Mintra Health, and your host for this series exploring the latest developments in optometry and ophthalmology clinical trials. Each episode, we’ll sit down with leading clinicians, researchers, industry experts, and innovators shaping the future of eyecare. From emerging innovations in study design, patient outcomes, and regulatory insights, we’ll take you behind the scenes of the clinical research transforming optometry and ophthalmology.
Whether you’re a clinician, researcher, industry professional, or simply passionate about advancing vision science, this podcast is designed to bring you thoughtful conversations and practical insights from across the ophthalmology and optometry community. Clinical Trials Spotlight is co-produced by Mintra Health and Ophthalmology 360 and Optometry 360. Thank you for joining us. Let’s get started.
In this episode, we’re diving into using AI in clinical trials. Today’s guest discussing this exciting topic is Ram Yalamanchili. Ram is the founder and CEO of Tilda Research, a company focused on building AI-powered teammates for clinical research organizations, sponsors, and trial sites. Ram is a successful entrepreneur who built and sold a company called Lexent Bio. After leaving Lexent, Ram looked around to see what other problems he could solve. He realized that the site to CRO to sponsor connection was the most tedious aspect of a study, and he believed that AI could fix this problem and relieve everyone’s burden, especially at the site. Ram then became a clinical research coordinator for 2 years to study the problem firsthand and then went on to build Tilda Research. Therefore, Ram brings a unique combination of deep technology expertise and healthcare innovation experience. Ram, welcome to Clinical Trials Spotlight.
Ram Yalamanchili:
Thanks for having me here, Andrew.
Andrew Pucker, OD, PhD, FAAO:
For our listeners who may not be familiar with Tilda, could you tell us a little about the company and what you’re trying to solve with the company?
Ram Yalamanchili:
Tilda is an AI teammates company. That’s how we describe our products. What that essentially means is we have agentic solutions where our AI teammates can augment existing ClinOps teams, and they’re able to perform work from an end-to-end loop perspective. We primarily focus on 4 different categories, I would say. First is TMF and regulatory management, so anything around documentation and managing the completeness and spectrum readiness and related workflows for TMF. The second would be site and study management. These are things where you have workflows related to site or study related vendors. Frequently, there’s a lot of communication burden involved in managing sites, and that’s an area where our AI teammates can come in and help do some of these workflows. Examples of these would be site startups, site maintenance, closeout, follow-ups around monitoring, querying, things like that. This is another category.
Third and fourth would be in the data monitoring and query management. Finance, specifically around optimizing payments to vendors and physicians, or I should say investigator payments. From there, there’s quite a bit of challenging problems. For example, reconciliation is an interesting and complex topic, especially in clinical trials, because you have a lot of invoice rules, which are typically non-standard and there’s quite a lot of work involved in managing that sort of a workflow today in the industry. We’ve built some AI teammates which can help optimize those workflows. Finally, forecasting, being able to forecast your outlay on a near-term and a medium-term basis, based on actual realistic information, which you’re seeing within your trial portfolio. Those are the 4 areas which we focus on.
Andrew Pucker, OD, PhD, FAAO:
It sounds like you’re making technology that can allow team members to focus on the things they really want to do, like seeing the patients or that sort of thing?
Ram Yalamanchili:
Yeah, absolutely. I think most of the workflows I’ve described focus on sponsors and CRO ClinOps teams. Certainly, when the workflows are inefficient, a lot of that burden gets translated back into the site workflow. To give you an example, we today power some of the largest CROs in the world, global CROs, and we see an interesting statistic when we benchmarked some of the workflows which contribute to site burden. A clear example, which I think most sites would relate to as ClinOps teams, like your CROs asking for the same documents or information over and over again, even though you’ve provided that information maybe as recently as just the day before. It’s an interesting persistent problem, which isn’t the end-all be-all, but it tells you the type of burden which gets introduced into the last mile of the research ecosystem when things aren’t as efficient as they should be right above it.
That’s just a quick example. What’s surprising to us is 30% of requests to sites, if you benchmark, is actually redundant workflow. If you look at all the communications made in a large cohort of sites and communication between CRAs or sponsors to the sites, around 30% of the time you’re basically asking for the information which you already have or the site actually provided you, but you’re just asking it again. For many reasons, maybe you had a churn in CRAs, maybe you’re starting an entirely new study, but there’s a different team in that study, but still the same sponsor, maybe still the same CRO. There’s a lot of information and knowledge which gets essentially re-queried and that leads to a lot of burden. Of course, that ultimately is on the site.
Andrew Pucker, OD, PhD, FAAO:
I love the idea of fewer emails. Also, it sounds like you’re making the sponsors look smarter. We don’t get the site’s burden with all these repetitive things that they hate doing.
Ram Yalamanchili:
Absolutely. I think the more exciting part about this, Andrew, is the compression and timelines and cycle times across different types of workflows. Historically, we’ve not had that kind of a benefit, but I think the most exciting part of what we’re seeing with our sponsors is cycle times compressing while you’re making the experience from a site perspective much better. It’s a real win-win on both sides.
Andrew Pucker, OD, PhD, FAAO:
I mentioned in the intro a little bit about you, but could you walk us through how you got from working at Lexent to Tilda just for background?
Ram Yalamanchili:
Yeah, absolutely. To give you a background about Lexent, I started at Lexent as a co-founder CTO, and Lexent was developing novel molecular assays. It was a whole genome sequencing-based assay in oncology. A lot of the interesting science when I was co-founding the company was around the bioinformatics pipeline and development of AI techniques to find the right signal within this very large dataset. Every time you sequence a patient’s blood, you get quite a bit of information and you’re trying to find some signal in there. I came in board from a machine learning, AI, bioinformatics side of things. Maybe about a year and a half into the company, what I realized is the rest of the time was mainly spent on complex clinical operations because we had to run large trials.
We were running global studies to power the data generation. A lot of the challenges which we were facing, I would say largely, were in the clinical operations space. That was a fascinating problem because I was learning for the very first time what it takes to build a biotech. Ultimately, the science was excellent. We had really good outcomes, and ultimately we got acquired by Roche for the commercialization. From my perspective, after the acquisition, I was convinced that one of the biggest challenges which has an unmet need right now is optimizing and helping solve for the ClinOps problem. How do you make that better? That took me on a journey into trying to figure out maybe we can do something on the site side.
The very first thing we did at Tilda was acquire a small site network in California. Myself and several of my product team started to go and work in these sites and experience the workflow firsthand as a coordinator. I think that was probably one of the most rewarding parts of our journey here because we learned a lot. We understand the nuances of what it takes from a coordinator to be a good coordinator, and it’s a very challenging job. It’s essentially, I would say, not given the type of credit which I think it deserves. It’s just an extremely challenging role where context switching is one of the most underrated, I would say. Also, if you think about it, context switching is actually not a core strength for most people.
We are not built for context switching. We mostly are focused on a problem at hand, and I think we want to be that way. We found a lot of interesting problems. Then when we looked deeper, what we realized is the solution may not actually be purely at the site level. You need to go a couple of steps above that. That could be the CRO or the sponsor, both or all of them. Of course, when we started looking into this, AI is a highly compelling solution. Right around the time when we were starting was when the first set of LLMs started to come out. Most of us have deep expertise in applied AI research. Coming from our NextGen background, we were able to transfer a lot of that into meaningful solutions here.
Andrew Pucker, OD, PhD, FAAO:
I totally agree with the clinical research coordinator position. Real challenging. You got to think about a lot. Really from the CRO side, if you don’t have a good clinical coordinator at your site, we probably don’t even want to work with you. You are as important as the PI, for sure.
Ram Yalamanchili:
Agreed, absolutely. If not more, right? That might be a controversial statement.
Andrew Pucker, OD, PhD, FAAO:
I was trying to not be controversial, but you’re probably right, I think. When we’re thinking about AI, I think a lot of people think it might be related to just going and recruiting patients. What do you think is really the greatest opportunity today related to AI in this space?
Ram Yalamanchili:
Yeah, I think what you point out is an important problem to solve. Recruiting patients, finding the patients. I sort of look at it as 2 separate distinct problems. The patient ID problem is different from the patient recruitment problem. You can identify through various means, and of course then you would have to bring the patient in and actually screen and recruit and randomize. We are not really, I would say, focused on the recruitment problem because I think one of the interesting and new things which are happening with the technology innovations coming out right now, especially with the rise of reasoning models and some of the more advanced techniques coming up on how the latest LLMs are, agentic workflows are starting to become reality or have become reality. There’s a lot of opportunity surface when you look at agentic solutions.
I think the recruitment problem largely is an information retrieval problem. It’s about sifting through large amounts of data, finding the right signal. Sort of what my background has been at Lexent, if you will, from a bioinformatics perspective or in that sense. However, I think where the industry has historically not had the right tooling or the innovative technologies to be able to make something like this happen is now possible. The clear opportunity from our perspective, from Tilda’s perspective, is that there’s enormous amounts of surface around how you can optimize workflow, the day-to-day work of actually performing the trial. Agent solutions are excellent for these kinds of situations. That’s where I think our focus has always been. It’s about building AI teammates, focus on the clinical research problem, divide it into the 4 pillars which we’ve spoken about, and build solutions for those.
Andrew Pucker, OD, PhD, FAAO:
We’ve talked about an AI agent a few times now. Could you define that for our listeners? I don’t think everyone’s familiar with that.
Ram Yalamanchili:
Yeah, absolutely. The way I explain it is if you think about how ChatGPT or Copilot work, they’re essentially conversational AIs. The way I think about it is you are having a conversation, you’re asking it questions, and it’s able to come back with reasonable answers. This has kind of been, I don’t know, I think ChatGPT has about a billion users monthly active right now. This is what the world’s exposed to as of today, largely. I think somewhere along the way, they say 99% of knowledge workers are on some form of conversational AI right now. It’s very prolific. It’s all over. I think where we are not yet seeing that type of a magic in the broad sense is what we are talking about, which is agentic.
When you chat, you’re asking for information, but what about asking it to do something for you? It’s able to accomplish it end-to-end. This could be something like, “I want you to go reach out to these 50 sites and manage a protocol amendment. There’s a substantial protocol amendment which just came out. Here’s what’s happened, here’s the context. I’d like for you, the agent, to go out and basically implement this across these 50 sites.” There’s a lot of nuances there. It’s not something which you can logically describe because the agent has to have some amount of intelligence on what does it mean to actually perform a substantial amendment. There is a lot of nuance.
As you know, in the research industry, for example, some sites might be in local IRBs, some might be under central IRBs. Each of those workflows have their own nuances. Depending on the protocol and maybe even at the site level, depending on which arm they’re on, there might be nuances there. There’s a lot of complexity which starts to come out all of these types of asks. Being able to imagine an AI teammate to do that has never been possible up until now. It’s just what the models and AI capabilities have evolved to a point where this is actually possible today. That’s how I look at it. It’s basically giving it a complex workflow, a task, but being able to do that in a highly accurate, highly scalable way.
Andrew Pucker, OD, PhD, FAAO:
I think that’s incredibly useful. You mentioned protocol as an example. We have a protocol change. We know that would trigger potentially an IRB amendment. Other documents downstream the protocol would need to be updated and filed at the sites. I foresee this helping you avoid missing things. I think in addition to it just being easier, you’re also going to be better at your job. I think that’s really helpful.
Ram Yalamanchili:
Yeah. I think one of the more fascinating things we are seeing and going back to the cycle time compression is the reduction in rework. Somewhere along the way, I think, as you were saying, the focus of our energies in building great ClinOps programs or bringing drugs to market has gone off into this area where there’s a lot of reworks and it’s acceptable. Everybody accepts it. When we did a benchmark on that in the global CRO footprint review part, it’s acceptable to do 40% rework in many of the categories.
For example, TMF I think is a great example; 40% rework. TMF itself is a lagging indicator. That’s not where the problem generally starts. That’s where it ends up. There are many upstream reasons why that’s the case. I see that as a very important metric which needs to come down; 40% leads to many other issues. Could be cost, it could be quality, timeline. I think even cutting that in half or a large percentage has a meaningful impact on the overall metrics and cycle times. That’s what’s exciting. I think that makes all of this really worthwhile.
Andrew Pucker, OD, PhD, FAAO:
Agents are helpful, but I think people are probably still important. How do humans fit in with this? Maybe you even want to give an example of how that might work.
Ram Yalamanchili:
Yeah, absolutely. I think the most important aspect of all of this is work is changing, but it is changing with humans in control. We live in an environment, we are in a regulated environment where governance is really important. AI governance is an area which we focus quite a bit on. We’ve done extensive work in coming up with the right models in terms of what it takes to have a robust AI governance policy and implementation within ClinOps states. We are not in a position where, or I don’t think the idea is to really say there’s no need for a human in the loop in this case. You absolutely always have to have that. Judgment ultimately will be the person who’s basically managing these AI teammates.
The way I look at the evolution of all of this is rather than doing the mundane work, which has a high probability of error and high probability of burnout, let’s have AI teammates do them. If you want to communicate with 50 sites to manage a complex workflow, let’s have the AI teammate do that work or coordinate that work for you. But you are the person where they come to you and say, “Hey, is this okay? I’ve done this much. Are you approving of what I’m trying to do?” In that sense, if you think about that paradigm, it’s a really, I would say, highly enjoyable way of working because you focus on things which are actually core skillset for you. You’re also able to do a lot more with the help of these AI teammates.
I think I’m also of a belief that the world is going to experience a massive shift in terms of how much work there’s going to be. I think there’s going to be an explosion in terms of number of clinical starts. There’ll be a lot more need for more experiments, more research. We’re seeing this across the pipeline. The early stage of the funnel, which is the discovery phase, is already one of the most funded, most interesting areas at the moment because there’s a lot of AI-driven activity and funding going into coming up with new molecules or new discoveries on that side. I think very soon, sometime this year to next year, we’ll notice is we just don’t have the capability to bring that many numbers of innovative molecules into the market and perform research and be able to do a good job.
We absolutely need a way to handle all this. I think because of that, we believe that AI governance plus being able to do more is what the world will move towards. One other example I have is in the organizations where we have established AI teammates into the workflow, what we have also noticed is teams generally grow. It becomes a competitive advantage. We have not seen any place where they’ve reduced that. It’s a very interesting phenomenon. It’s the counterintuitive thinking in reality in what’s happening. We’re big believers in saying the enablement of these kinds of technologies will lead to an explosion amount of work which is required and we need more people because of that.
Andrew Pucker, OD, PhD, FAAO:
I love the idea that AI is making new projects to study. It’s making it easier for people. Their jobs might be a little different, but all of that together is probably making more jobs overall.
Ram Yalamanchili:
Absolutely. We already have anecdotal evidence of that within our customer base, but I think if you, I think extrapolate that, I see a trend there which is quite compelling.
Andrew Pucker, OD, PhD, FAAO:
In the research space, especially when you’re doing FDA trials, we know that it’s highly regulated. How have you navigated that hurdle with your technology?
Ram Yalamanchili:
Yeah, the FDA is actually quite AI forward, I would say. We’ve seen several examples where that’s how it’s proliferated. One area of the FDA has put out guidelines is how to build AI-based systems and what does it mean to bring in AI into code corporations? I think the key here is that you need to have an expert in the loop. That’s where the human in the loop comes in. The FDA is not okay in having AIs do work which has no supervision, no governance, and completely on autopilot. I think there’s even been cases recently where they’ve come down hard on a couple of examples where protocols were written through AI and not reviewed and there was no oversight.
Building the right solutions means you have to work within the framework of what the regulatory allows. Today, that means you have to have excellent AI governance-based functionality in your product. One of the other areas where we focus on when we build these AI teammates or even the platform in which you work with our AI team meds, it’s called Tilda Sense, is the AI governance module. There’s a lot of knobs which we provide. Based on the customer’s AI governance SOP, you have to set it up in a way where there’s adequate human oversight into what’s happening on the AI side. It’s all, of course, altered, documented, and so on and so forth.
Andrew Pucker, OD, PhD, FAAO:
This is kind of related to my last question. What potential questions do you often get from sponsors related to this technology? I’m sure one of them is related to regulatory, but is there any other things that you worry about and how do you mitigate those issues?
Ram Yalamanchili:
Yeah, absolutely. I think the very first question is typically we show a snapshot of how this works, let’s say a demo. We generally have a reaction where, wow, this is incredible. Does this actually work? How real is this? That’s an excellent question to ask because I think one of the great things about these AI tools is the ability to create a very quick demo or something which sounds reasonable or looks interesting and excellent has really come down. I’m sure you’ve seen apps where you can wipe cord something and then it looks very, very reasonable. I think there’s an element of trust and disbelief, which we need to overcome when you bring in these technologies and for the right reasons. Sometimes we even educate the customer. We say, “Hey, these are the right questions you need to be asking.” We’ve actually documented this.
We have published on our own website, what are the ways you would want to buy AI solutions? Because we understand that we all live in a very tight-knit community and ecosystem and demos and AI which is not controlled or is not quite aligned to the mission can actually destroy the trust in AI itself. If people start adopting tools which do not have that breadth and depth of the reality of what it can do, can it actually perform the work on a consistent basis? Is it going to lose alignment? By that what I mean is today it’s working well, but tomorrow it may not work as well. A good example of something like this is what we call drift. If you, for example, look at ChatGPT or a Copilot example, let’s say you asked a question today at 8:00 AM and at 8:05 [AM] you ask the same question again. It’s never going to give the exact same answer because these are stochastic processes.
It’s going to be about the same, hopefully. Hopefully, it’s right both times, but there is no guarantee. There’s an element of drift in terms of every time you query it, it might actually give you a different answer. A lot of our work goes into making these AIs extremely consistent, building the science behind how do you bring applied AI research techniques to make our AI models quite consistent, high quality, and also keep that alignment over time and monitoring all of this workflow across many, many customers and many, many studies and things like that. Then you really go into this questions of, “Okay, show me the advancements you have in monitoring these AIs. How do you know if these AIs are continuously performing what they’re supposed to be saying or doing? How do you know they haven’t lost a certain part of my training, which we’ve done, we’ve spent all this effort, but then they forgot all of that in the next iteration of an update?”
There is a lot of nuance and challenge and complexity in maintaining these AI-based teammates from a developer perspective, from our perspective. It’s a mix of educating the customer on these are the right questions to ask. These are the things you want to evaluate these technologies on. Then gaining that trust and the ability. The second question, which also I see it, but if we don’t see it, I make sure I say, please ask us this question. How are we going to make your teams AI-fluent? Because just giving a technology at this time, especially to these kind of AI teammates, is not a recipe for a success in deployment.
What you really need is you have to invest in your team to make them AI fluent because that’s absolutely a requirement. I almost think of it as like if you care about your organization and your team, you want to make them successful in using AI tools and you want to make them as good as possible in a very quick term. That’s been a second big part of our focus. I know we spoke mostly about our AI teammates and technology, but if you look at what we’ve done over the past year, built a lot of coursework around what AI fluency means for the clinical operations space. We have a free masterclass on our website. It’s got quite a bit of traction. For all of our customers, we are very focused on bringing training. That’s a big part of our focus and the way we build our organization. I think between those 2 is where we generally see a really good balance.
Andrew Pucker, OD, PhD, FAAO:
That makes total sense. You need to put a smart prompt into the machine for it to give you quality data. You’ve talked about your masterclass. Are there any other things that maybe a clinical research coordinator or someone like that should be trying to do to get better at this so they’re ready for it?
Ram Yalamanchili:
Yeah, I think I would recommend not just our masterclass. I think our masterclass is mainly focused on understanding AI from a perspective of a clinical operations, so it’s coordinators, sponsors, CROs, anyone in these roles, a very high level generalized format. I would encourage, there’s several course works around AI, better prompting, prompt engineering. I think Udemy has several good courses. I think even online outside of what Tilda has, I think there’s excellent resources around just general-purpose AI. I do think being able to understand how to prompt is probably one of the most important skill sets to be developing right now.
Of course, depending on whether you’re a user or an implementer, there’s different threads. For example, if you’re implementing or purchasing AI-based solutions, then you go back to what I just said, how do you evaluate these systems? What is the right metrics? What are the right dashboards you want to see from our technologies or any other technologies? I think I look at it as if it’s a user, then prompt engineering is really critical. If you’re an administrator or a implementer of these technologies, then there’s other specifics about the technology itself, about the AI itself.
Andrew Pucker, OD, PhD, FAAO:
I think your answer is yes to this question, but I’m going to ask you anyway. Do you think that in the future, all clinical trials will have AI embedded in it? Then how long do you think it’ll take us to get there? Because it seems like this technology’s evolving really quickly.
Ram Yalamanchili:
Yeah. No, the answer is yes. I actually think it must be a yes because the expectations on quality, time, and cost are changing rapidly because we’ve always historically had a specific curve on all these 3 metrics. If you look at our current implementations, current footprint, we’re coming in at a reach into the industry where it’s becoming quite ubiquitous. By the end of the year, we’ll have tens of thousands of sites powered by our AI team. Tens of thousands of sites related workflows powered by us. That’s getting to a point where you’re a substantial majority of the market, over 50% of the market will basically be adopting some form of an AI teammate to perform this work. It’s a very powerful position to be in. It says a lot in terms of why we’re doing it and what’s happening in the industry. I think I expect pretty much everybody to have some kind of a requirement to build these into their trial ops.
Andrew Pucker, OD, PhD, FAAO:
That’s incredible. That’s even going faster than I thought it would.
Ram Yalamanchili:
Yeah, it’s absolutely. We are in a once in a, I think it’s a generational shift in terms of what’s happening right now, especially given the pace at which adoption is happening at many of these organizations.
Andrew Pucker, OD, PhD, FAAO:
Related to that, do you think we’re going to get drugs to market faster? I know we need to have clinical trials run for a specific amount of time, for example, to look at efficacy of a drug. How do you think your technology will expedite trials?
Ram Yalamanchili:
Yeah, I think that’s a loaded question, Andrew. I think there’s an element of optimization on cycle times, which our AI teammates can provide, but there’s also several areas where I think we as an industry need to do better from timelines and managing those timelines. I mean, you have regulatory requirements, there’s filings, which take a certain amount of time. I think there are areas where you can control these kinds of cycle times and there’s going to be areas where we cannot control it, at least as of today. To answer your question, can we actually reduce the timelines on the overall studies? I’m very optimistic. I think it will happen. It is going to happen. However, by how much is going to be a conversation, not only from a technology perspective, but also from the regulators and some of the other ecosystem in this industry.
We have to look at it from that perspective. The other, I think you mentioned is, for example, safety and follow-up timelines, things like that. We don’t particularly have a focus on doing that type of, I mean, meaningfully impacting that timeline or trial design. I think if you say, “Well, we have better in silico modeling for safety or for certain types of information which we would otherwise not have,” I could assume that that could translate into better timelines and maybe different types of trial design, but it’s still early and I think I’m following that space very closely. It’s fascinating what’s happening in that space from others in the industry. I’m an optimist though, so we’ll see. I really do think we’ll have an impact on the timeline. We have to.
Andrew Pucker, OD, PhD, FAAO:
I hope so. We need to get treatments to certain patient groups fast because they have no options at all. I think it’s going to be a partnership with the regulators and technology people and clinicians. Right now, you’re making trials better, safer, and more efficient. Hopefully, like you said, eventually that also translates into faster to market. Going with that, what are you most optimistic about clinical research going forward?
Ram Yalamanchili:
I think the most, at least what I’m seeing is from my vantage point, the traditional rep of clinical trials being highly burdensome, I think that’s going to change meaningfully. I think it is changing meaningfully in areas where we are working, and that’s a great thing. I’ve not seen that since I started this journey of managing clinical trials and working as a coordinator. Today, of course, being in the mix of product and delivery into some of the largest clinical operations teams out there. I think that question of burden, whether it be site burden, it could be ClinOps burden at the sponsor level, at the CRO level. I’ve been hearing of this particular word of burden in every conference I’ve been probably to, and I’m sure you have as well. I actually feel it will be evident that we are making a dent there, and that’s awesome. That’s actually quite great to see happening.
Andrew Pucker, OD, PhD, FAAO:
Great. I agree. That’s awesome. Keep doing the good work. With that, I’d like to thank our guest, Ram, for joining us on the podcast today.
Ram Yalamanchili:
Thank you, Andrew. Appreciate being here. Great questions, by the way.
Andrew Pucker, OD, PhD, FAAO:
I also would like to thank you all for joining us today on Clinical Trials Spotlight. If you enjoyed today’s episode, be sure to subscribe and share the podcast with colleagues and others passionate about advancing ophthalmology research and patient care. Thanks for listening. We’ll see you next time.
Related Content
