The Art of Simulating Biology with Certara

Speaker 1:
Welcome to TD Cowen Insights, a space that brings leading thinkers together to share insights and ideas shaping the world around us. Join us as we converse with the top minds who are influencing our global sectors.
Brendan Smith:
Welcome back to another episode of Machine Medicine: AI & Health Care, TD Cowen's podcast series where we bring you the latest and most important takeaways from the state of AI in the healthcare sector today. I'm your host and TD Cowen Healthcare Analyst, Brendan Smith. And today I am joined by Certara's Chief Executive Officer, Bill Feehery. Bill, it's great to have you and welcome.
William Feehery:
Great. Thanks to be here.
Brendan Smith:
Yeah. So, for anyone new to the podcast series, Machine Medicine aims to break down the use of artificial intelligence across healthcare into bite-sized digestible points, one episode at a time, highlight the biggest misconceptions, and then recontextualize each piece back into the bigger picture. And today, Bill and I are looking to check under the hood of Certara's biosimulation engine. And discuss not only how computational modeling fits into the rapidly evolving field of AI, but also how it's already addressing some of healthcare's biggest bottlenecks and where the technology can go from here. So, let's just dive right in. First, maybe let's start with a quick definition of what we actually mean by a biosimulation model and how it differs maybe from some other computational approaches. And realistically, I guess, how an investor can wrap their head around just the building blocks that make up Certara's platform.
William Feehery:
Yeah. Thanks. It's great to be here, Brendan. That's a great question to start with. So, in a nutshell, biosimulation is the use of computer models to simulate biological systems. And what we're doing is we're combining biology, mathematics, and computer science to create, let's call it a virtual representation. Some people call it a virtual twin of how a cell, an organ, or even the entire human body functions. So, the point of all this is to use those models to predict how a system like the human body is going to respond to a drug or a disease. And the ultimate purpose, the value in what we're delivering, is to reduce the need for clinical trials, to speed up development or sometimes it's just to be able to ask what if questions. How does this compare with different types of treatment, for example, different molecules I might consider?
So, to come back to maybe your question about how all this comes together. We've been working on this for about two decades right now. We've been building it up slowly from the known biology. We've had a lot of success. A lot of drugs have used our technology today that have been approved. And it's getting better and better as we go forward.
Brendan Smith:
Yeah. And I mean, I think, look, that's a great intro to a lot of the topics we want to cover today. And computational modeling, of course, itself does not inherently indicate the use of artificial intelligence, which is frankly the broader focus for the whole podcast. But it is frankly, look, I think a common misconception amongst some people who are trying to get their feet wet in this whole space. So, I guess for someone on the outside of all this, how do biosimulation and systems modeling in general fit into the context of AIML specifically? And I guess maybe more to the point, how is Certara looking to leverage AI across your own platform?
William Feehery:
Yeah. That's a great question. So, you're right, we at Certara have been working on biosimulation for literally over 20 years now. So, it's obviously long before AI has reached its current level of excitement and sophistication, but it's also given us a lot of time to really prove our technology. So, one thing I should emphasize is biosimulation that we're doing at Certara has truly proven there are... we've counted over 100 approved drugs with hundreds of labeled claims, which we can trace to the use of biosimulation. And what's important about that is that it's enabled companies to avoid some level of clinical trials, which is the most expensive part of drug development, by using our software. So, coming back to AI. The way we see is AI doesn't replace biosimulation, but it offers some really intriguing opportunities to enhance and extend what we're doing in biosimulation.
So, we think that they're going to work in concert as we go forward. So, specifically at Certara, we're really excited I think right now in the near term about two areas. So, one is that we can use this idea of AI to supercharge biosimulation. So, think about this, to create biosimulation models, our scientists have to do a ton of work to understand and synthesize all this in all the papers of the known biology. And then we have to find a ton of experimental data, sort through it, figure out what's important, normalize it, and validate those data. AI can do a lot of that much faster, and it can enhance the speed, and basically the productivity of these experts that we use.
So, we're looking at tools right now that basically let us process a lot of the scientific literature, that basically, they can look at a lot of data and pull out the key pieces we need. And the idea here is we want to move biosimulation away from this group of super experts, which frankly, there just aren't that many. And we want to expand both how productive they are, and also make it so that other people can become fully conversant in the use of biosimulation. So, the second thing we're looking at doing, because I said there were two, is we're looking at how we can use AI to incorporate more data into biosimulation. Biosimulation's main limitation is we try to model the known science, but there's lots of things out where you know it's important, but you don't necessarily know all of the mechanisms why.
So, for example, omics data. You know it's important. Statistically, you can pull out populations, but we don't necessarily know the specific chemistry that an omics combination really results in. So, in the past, we didn't model that in biosimulation. It was a bit of a limitation, but now we can combine it with AI. We can do things like pulling it, using AI to define certain subpopulations. And then we can use biosimulation to model what's happening in those subpopulations. You can go a lot faster. And I think these hybrid AI biosimulation models are really going to be the way to the future.
Brendan Smith:
Yeah. I mean, again, this is a great segue. I think even just beyond where you guys already are today, this extensive end-to-end in some respects, biosimulation offering that Certara has, I mean. But when you look broadly across the entire healthcare sector in general, I guess, where do you see some of the most tangible, maybe near-term opportunities and use cases for AI and ML, even just within maybe the spaces that you're already operating in, but also just where this technology realistically could go for healthcare?
William Feehery:
Well, I think somebody said earlier today at your conference that in a couple of years, every company will be an AI company. We'll all be using it across our entire careers. But if you look at healthcare, I think there's a lot of companies out there trying to solve the discovery problem, which is just how do I pick through the infinite universe of all the possible molecules that I could make into a drug and narrow that down quickly. We've been more focused on using AI in a clinical setting. So, when a drug gets into a large population of people that vary by genetics or by age, weight, or maybe comorbidity, other diseases that they have, how's it going to really react when it gets out there "into the wild?" Other people are looking at, and we're doing some of this as well, real-world data. So, instead of doing clinical trials, can we pull out of the data we need from real-world medical information?
So, I think those are all legitimate places this is going to be. In the really short term, I think what people are really pushing in the short term, we have, I'd say two tools that we're really very interested in. So, one is it turns out scientists don't like to write any reports. They'll edit reports. But what are we doing in biosimulation? Well, you do a study and then you have to write a report. That report might go to a regulator who's going to read the report. And we can accelerate that entire process using AI. So, we've launched the tool in that area. Another piece of it is these are very complicated models. They're non-intuitive to humans, despite the fact that I've said we've modeled the underlying science. They're differential equations. They behave in non-intuitive ways. And so, part of biosimulation is not just doing the modeling, but explaining what's going on afterwards. And it turns out AI is really, really good at doing that.
Now, these things might seem, I don't know, a little bit more mundane than the way you hear in AI. But if we can take a lot of the work out of biosimulation and we can expand the number of people that can use it, our vision is that we can push biosimulation and many more drugs that probably should be doing this, but for various reasons haven't over time. So, it kind of expands our market. And it spans the vision we have at Certara, which is how do you take the drug development, which is really a very low probability endeavor, and really supercharge the probabilities? How do we do this better?
Brendan Smith:
Yeah. And bend the curve. I feel like we constantly are hearing this expression, how am I going to bend the curve in the right direction? I mean, you already touched on what some of the areas that biosimulation is really able to capture, share, and is particularly well-primed, whether that's different therapeutic areas, drug modalities, different phases of drug development. But I guess really where you see what Certara is able to do today, what your platform offers versus very, very traditional drug development approaches. I mean, I guess I'll have to caveat this by saying by the time this podcast is even out, we'll probably see even higher percentage of companies already leveraging some of these offerings. But if we were to rewind the clock to 10 years ago, a very traditional standard way to develop a drug versus leveraging any one of Certara's platforms, I mean, where are the best areas that you feel like biosimulation outperforms versus what's been done in the past?
William Feehery:
Well, that's a good question. Biosimulation excels in explaining how a drug matches the known science. And that can be really important, because drug development at the end of the day has been a statistical science. We do a placebo-controlled trial. We do some patients with, some patients without. We look for a statistical significance, which is a great thing to do. But I always remind people that a 95% confidence interval means you're wrong 5% of the time. So, really, you need to couple that. And scientists know this. They couple that with the underlying science as to does my trial actually match what's going on. And we enable that. And by doing that, we can do some really interesting things. So, we can predict what's likely to happen with the known data in the next trial you do.
If you can predict what's likely to happen, now maybe were not 100% right, maybe we're partly right, but you still might change what you do in that trial. You might make it smaller. You might not do it. You might include different patients or exclude different patients. You might give a different dose. So, the ability to change the path of drug development, that's really the heart of what we're talking about in biosimulation. It makes it hard when people ask me, "Hey, what's the value of this?" Because I'm comparing someone who used biosimulation to drug development the right way with some path that they didn't already take. But I think one of the signs of success is just the number of drugs that are really using this right now. And I think that's going to grow considerably both in a number of drugs and also the use cases within that drug development as we go forward.
Brendan Smith:
Yeah. I mean, think to that point, I think you guys have cited this a few times. The upwards of 90% plus of drugs that have been approved over the last 10 years have used or leveraged some form of Certara's software in some capacity. And it's like that level of entrenchment, I think is continually underappreciated when people are thinking of AIML biosimulation in general as like, oh, this is around the corner. It's here. It's been here.
William Feehery:
Yeah. And I think that's one of the things that people sometimes don't think about when they are looking at AI and drug development. So, a lot of people are talking about how do I pick a drug better. You absolutely have to do that, but we really want to do the clinical trials more efficiently or as fewer of them as possible. I mean, they're really expensive. You're giving an unknown substance to a human. You don't want to do that any more than you have to. There's a potential for unexpected outcomes that potentially could kill your drug. So, that's really where a lot of the cost of drug development is. It's in how do you recruit all these patients and track them potentially for years. If you can do a lot less of that, you can pay for a lot of biosimulation.
Brendan Smith:
Yeah. Yep. And I think I would also be remiss if I didn't ask you a little bit about FDA. So, I mean FDA is notably shifting its priorities really when it comes to integrating, and validating, and looking to incorporate a lot of these AI computational modeling techniques. And has really highlighted biosimulation as one of their preferred favorite alternatives to some of this traditional animal testing that we've seen historically. So, I guess, first, for you and for Certara, how has the recent guidance and messaging out of FDA impacted your business? And where are you seeing the most new customer interest in terms of new current platform offering?
William Feehery:
Yeah. So, we've gotten a tremendous amount of interest in this. This has been a long time coming. A lot of people in the biosimulation community have been pointing out that the data that you get from animal trials is not always particularly useful in drug development, to be honest. The FDA is aware of that, acknowledging it, but they're the main regulator around the world. They have to send a signal to people what they want to have forward, and they've done that here. I think there's going to be multiple steps to this. There's plenty of examples where we use models today to effectively eliminate animal models. I'll give you one example, which is in cell and gene therapy. I mean, the FDA for some time now has accepted that you're going to use a model, because it really doesn't make sense to give a cell and gene therapy to a non-human if it's designed for a human. Monoclonal antibodies that the FDA cited is another good place to go.
But I think there'll be a couple stages. Animal trials are used for several different purposes. One of them might be dosing, which I think for a lot of modalities we can handle now. Obviously, toxicity, we can handle some of that now. We could probably invest in more. There's probably a few cases where the people are doing long-term mutagenic studies, things like that, where it'll be probably hard to eliminate all animal trials. But I'm not sure it's really the FDA's point in pushing this. So, I think it's great. I think it's obviously a good idea to reduce the use of animal models from ethical standpoint, from a cost standpoint. But frankly, just if we can get drugs developed six months earlier, that will benefit the people around the world.
Brendan Smith:
Yeah. And I think that's kind of a point that is sometimes lost in the, just say emerging noise from a lot of this space. It's not necessarily that every drug is going to be perfect or every drug is going to work. It's more that if the drug is going to fail, it's going to fail faster, hopefully cheaper, and with better resources that went into actually finding that ultimate conclusion. And to your point, out of FDA, I mean next steps then become the very next question out of people's mouths. So, I think they've intimated that they're going to have a pilot program, will be an important next step for them, introducing some new roadmaps for each of the technologies that they've highlighted here.
But it should also help demonstrate the ability of these new alternative methodologies or NAMs that they're calling it. And how they perform relative to some of these traditional animal testing, which I think will go a long way to get a lot of people on board. So, I guess, how do you think the FDA will look to fill those pilot spots? And I guess, where do you see Certara fitting into that conversation, as people maybe outside of the space are trying to decide which different offerings now to dip their toes into?
William Feehery:
Yeah, that's a good question. So, the FDA has, I think, a track record of doing what you're talking about, having a pilot study with a group of companies. We're looking forward to seeing what the FDA decides to do there. Our guess is the most interested companies will be some of the larger companies that are doing lots of trials, where there's potentially both the sophistication and the benefit to push this forward. We think that specifically related to monoclonal antibodies what the FDA come out with, we've got a track record and biosimulation models in that area we've worked with quite a number of customers on. So, we think we're likely and happy to provide support to those companies as they go forward. But then, hopefully, that expands quite quickly, and passed that pilot group, and that success leads to even more opportunities to reduce animals.
Brendan Smith:
Yeah. I think this is honestly critically important for a lot of people who want to understand the actual impact of the guidance coming out of FDA and get a little bit better sense of just actual what timing looks like. Nothing happens at the federal level overnight. The gears of federal government grind very slowly, even in the best of times. But I guess, to that point, and in keeping with that aspect of the conversation, what do you feel from where you stand now and how things have evolved recently? What do you think the investment community maybe under appreciates or just misunderstands altogether about biosimulation? And maybe just what's something in that respect that you think is really essential for anybody, any investor in 2025 to really understand?
William Feehery:
Well, I think you hit on it just a second ago when you talked about the speed at which some of these things work. So, we've been around for some time. Certara, one form or another, has been working on biosimulation for almost two decades. In the very beginning, it was tough going. There was just a few people that believed in it. The regulators didn't accept it. It didn't work as much as it does now. But we've had quite a bit of time where year after year we're building up more sophisticated models, validating, and taking the regulators, getting actual experience with approved drugs. And that's been building up at the same time that the biology, the known biology has been getting more sophisticated. The available computational resources and data to us are way larger. And now AI is coming along.
So, I think that we're right at the cusp of a real renaissance, I think, in biosimulation. We're at the point where this is proven enough. It would be really unusual, I think, now to develop a big drug that used no biosimulation. But there's a lot of places where people could use this, where we're still putting the models together, and there's going to be a lot more as we go forward.
Brendan Smith:
Yeah. I mean it kind speaks to that 90% number that you guys have thrown out. It doesn't necessarily mean that all 90% is using equally or as deeply as they realistically could. A single touch point will technically get you in that, but that doesn't mean that somebody who has 100 touch points will actually get there faster and maybe more successfully to that same point. All right. So, look, I know we've covered some great ground today. And it's a conversation I'm sure you and I will continue to have over the months ahead. But before I let you go, there's one thing I do like to ask all of our guests. And it's if everything we've discussed today maybe goes over someone's head, but they've made it with us this far, what's one point you would really want everyone listening in to remember and just take away from the conversation?
William Feehery:
Yeah. I think the one point I'd like to say is drug development is a very low probability endeavor. I said that earlier. So, I would say conservatively, if you look at this, 90% of drugs that people start after discovery fail. And it doesn't have to be that way. If we're smart about using modeling, and biosimulation, and pushing these tools forward, we can change that by quite a bit. And since we're modelers, we've done modeling here. And we've shown that even fairly small changes in those probabilities can really change the cost of drug development. And if you can change the cost of drug development, we can basically develop more drugs, which benefits the whole world. So, I guess the thing to mention is the low probability statistics of this industry, it doesn't have to be that way. And I don't think it will be as we go forward.
Brendan Smith:
Yeah. So, that's great. And I think with that, I do want to thank you for hopping on and talking us through what really is this cutting edge of this marriage between software and healthcare tech innovation. I'm sure we'll have plenty more to discuss in the weeks and months ahead as it pertains to all this. So, thanks very much for joining, Bill.
William Feehery:
Thank you.
Speaker 1:
Thanks for joining us. Stay tuned for the next episode of TD Cowen Insights.