3148: AI Meets Pharma: How BioPhy is Accelerating Clinical Trials

[00:00:03] What if artificial intelligence could transform the complex and costly process of drug development and reduce failure rates which will lead to bringing life-saving treatments to market faster? So today I'm joined by Dave Latshaw, CEO of BioPhy. And with his wealth of experience in AI and machine learning, Dave is now leading a revolution in AI-driven drug development

[00:00:30] focusing not just on discovery but what happens downstream where clinical trials, regulatory compliance and real-world healthcare outcomes intersect. So I want to learn more about BioPhy's AI-powered platforms and how they're making waves in the pharmaceutical industry tackling challenges like scalability, stakeholder adoption and compliance automation.

[00:00:54] So the big question, how is BioPhy leveraging cutting-edge technology to make a real-world impact on healthcare outcomes? Well, enough from me. Let's get Dave onto the podcast now to find out more. So a massive warm welcome to the show. Can you tell everyone listening a little about who you are and what you do? Of course. Happy to be here. My name is Dave Latshaw and I'm the CEO of BioPhy.

[00:01:23] So BioPhy is focused on artificial intelligence for drug development. I say that very specifically because there's a lot of focus and hype around AI for drug discovery right now and we're focused on opportunities downstream from that. And Dave, with your extensive background in AI and also to mention as well the application in life sciences, I'm curious from what you're seeing out there because there is so much hype that surrounds AI.

[00:01:51] How do you see or believe AI is going to reshape things like drug development in the processes that you use? And what distinct advantages do you think AI-driven approach can offer over those traditional methods? Because we often hear about, hey, this is going to be great for that. It'd be great to compare and contrast how those two worlds might look or might be different. Definitely. So the way I think about this is sort of timescales and where the impact is going to be.

[00:02:20] So if you take the drug discovery and development process, you know, typically you would start out with your discovery work, then move to preclinical, clinical trials, commercial approval, and then, you know, the post-approval commercial activities. The best way to think about, at least how I believe the value is going to be realized and where you'll see the vast majority of deployment is to work. Deployment and success is to work backwards.

[00:02:48] So if you think about the reverse of that process, starting with the post-approval commercial activities, then approval, then clinical trials, then preclinical, then discovery.

[00:03:02] I think that's actually the order that we're going to see the vast majority of success and value realization in, in part because it takes a lot longer time to realize, you know, these larger speculative investments in the discovery phase. But two, the immediate ROI and incentive for investment in those later pieces of the process is very high and people want value today.

[00:03:31] And it's, you're not really going to find a whole lot of immediate value delivery in, you know, a artificial intelligence platform for drug discovery that has seven years until the first, you know, clinical candidate ends up being tested. Love that. And I think it's probably a great opportunity to introduce everyone to BioFi's research, microbiology consultancy, and some of the work that you're doing there.

[00:03:58] So can you tell me a little bit more about the proprietary operational assessment model that you have there, how that integrates scientific, clinical, and regulatory insights to ultimately predict the success of clinical trials more accurately? Because it feels like quite a big moment with what you're doing here. I appreciate that. So the thesis we have here is that there are a few really critical decision-making processes in the drug development lifecycle.

[00:04:27] One of them is very simply to pursue the right programs. And if you do that, then you can reduce a lot of the friction and wasted capital in the drug development process.

[00:04:39] So one of the pieces of technology that we built is geared towards integrating that data comprehensively, the biological, the chemical, the operational, in order to get a perspective on not just kind of the absolute level of success. Somebody should expect from a program like how is this clinical trial going to perform, but relative opportunities.

[00:05:05] If you're operating in a disease area that is very challenging, undoubtedly the expected success rate is going to be relatively low. But even within that, you're going to want to know in a risk-adjusted way how your program is going to outperform others. And that really helps from a competitive landscape perspective and trying to understand where your positioning is in the market. So that is one of the kind of initial technologies that we knew would be sort of fundamental to what we were building.

[00:05:35] And along with that, we've built kind of a regulatory and quality engine as well, which is oftentimes referred to as BioFireX. But at this point, everything has evolved where everything is in a single platform now. When we work with our partners, the availability of both of those solutions is through the same piece of software.

[00:06:01] And it allows them to get insights not only into their own portfolio and competitive landscape, but also start looking at how do we manage and optimize internal operations, especially against the changing regulatory landscape externally. And I would imagine that the journey from AI theory to practical application in drug development is incredibly complex. So what challenges have you faced while translating some of these cutting-edge AI research into operational solutions?

[00:06:30] Because I imagine it's not as straightforward as we'd all like it to be. I'll give you one operational and one non-operational, both of which I'd consider major obstacles to any adoption. On the operational side of things, I would say something people underestimate is scalability. With the tools available today, it's pretty easy to go from I have an idea to here's a prototype in a few days.

[00:06:57] And that gives people, I think, a false sense of what they can do with the technology without additional resources and consideration. Building a prototype is one thing. Building a company-scale solution is completely different. One thing that often doesn't become apparent when you're building these sort of technologies until you're actually doing the work is how much it actually costs to run the software at scale.

[00:07:27] Only then do you start to understand why being efficient with the compute, with the data, and those sorts of things becomes a major, major consideration. Especially if you're a pharmaceutical company trying to build something in-house. For startups, which are kind of subsidized with additional capital, they might be able to absorb some of those losses while they grow. But it's a very challenging thing that I think people often underestimate.

[00:07:53] And the financial of companies can suffer and create extra challenges where maybe they didn't anticipate them. Now, on the non-operational side, frankly, the biggest barrier is probably something that's not specific to AI, but is with all new technology, which is just adoption.

[00:08:15] How do you work with your customers, with your stakeholders, and get them comfortable with something which is brand new and cutting edge? Especially because there's typically always some sort of background fear around job replacement, potential downsizing, cutting of job responsibilities, and those sorts of things.

[00:08:40] So that part is a bit of an art because I think it's a little bit more psychological in how you approach the way you articulate the software to be used as well as how you plan to work with it. So I think those are probably two of the biggest ones that are going to be kind of universally experienced if you expect to do anything at reasonable scale and impact.

[00:09:04] And just to help anybody listening try and understand how this technology can make a real impact and a real difference here, especially with some of the work that you're doing. Can you tell me a little bit more about your AI engine and Gen AI solution and how these technologies specifically function to better enhance the outcomes of clinical trials and reduce failure rates? Because it's not just about hyping up another AI engine and Gen AI, etc. It can make a real difference here, right? I think so.

[00:09:33] And one of the things that I'm always really adamant about when I speak to people about what we're doing is actually just not just what we're doing, but what the industry is doing as a whole is people should not just be saying generative AI is this new thing. We need to have a budget for generative AI. Go find generative AI things to buy. That's a bad way to do business, and that is undoubtedly going to lead to a lot of lost time and capital.

[00:10:02] What people should be doing is asking with the capabilities of this new technology, what use cases are now possible that weren't before and looking in that direction rather than just forcibly saying we need to invest in generative AI.

[00:10:18] So because of that, the advent of or maybe popularization of that technology, because the actual technology has been around in some form or another for quite a time, is being able to work more extensively with written language and its interpretation.

[00:10:42] That opens a lot of doors when it comes to clinical trials because every clinical trial is unique. Otherwise, you wouldn't be running it and you could just reuse the results from previous trials as evidence. And using that information as a way to enhance other predictive technologies is kind of paramount to success, especially on the operational side of things.

[00:11:09] Additionally, I mentioned some of the work that we're doing in the regulatory and quality space. That is almost purely a language and text based function within drug development.

[00:11:24] These are the people that are responsible for ensuring that the way a company operates is in line with expectations from regulators, in line with internal business policy and being able to actually execute on those things. So they have to manage a massive amount of documentation and policy that dictates those things.

[00:11:50] Now, as you can imagine, if you have 100,000 operating procedures and you have to constantly monitor that and ensure that it's up to date, both with internal and external expectations, that's a huge job. And that's exactly why companies have entire departments full of people that are responsible for exactly that.

[00:12:09] So we believe that creates a very large opportunity to take computational approach using the technology behind generative AI to solve the problem of effectively automating the process of achieving regulatory compliance, but with people in the loop.

[00:12:29] I do believe for pharmaceutical companies in general, the vast majority of these use cases, generative AI or not, always, always start with a human in the loop and then progress and mature from there, especially because of the regulatory considerations around how they are viewed and the increased risk associated with removing humans from the process.

[00:12:56] And before you came on the podcast today, I was doing a little research on you. And one of the things that I learned quite quickly is this is not your first AI rodeo. And during your time at Johnson & Johnson, I read that you'd worked on implementing AI solutions that made a big impact on global health outcomes, such as the COVID-19 vaccine rollout. So I've got to ask, how have those experiences influenced your approach and vision at BioFi? Massively. So in a few ways.

[00:13:25] So I'd say on the technology side of things, there were a few things. One, getting the experience around how to build teams around the development of technology and then simultaneously manage business exposure and what you need to do in order to get adoption for those things with an organization.

[00:13:54] And scale it are pretty paramount. It also gave me the ability to learn the process of drug development end to end. And when you do that and you work in a role that is based specifically around technology, it's very eye opening because you can see clearly where the organization is spending their time developing solutions, where there are already things in market, where the gaps are.

[00:14:20] So a lot of what we're doing is geared towards my learning and understanding of where I believe there's opportunities that are underserved based on what I saw from my time at J&J. So there's a lot of lessons that I learned among those things.

[00:14:39] And maybe more personally, one of the really important things that I learned doing that is for me being close to the impact and the creation of the technology was one of my major motivators. That was what really got me up in the morning. And that is what kept me going through the time at J&J, knowing that what I was building was having a direct impact on people's lives.

[00:15:06] And that was also one of the major drivers of me being interested in starting BioFi with my co-founders, because I knew that that would bring me back to about as close as you can possibly get to having to build that solution and being directly tied to the impacts.

[00:15:26] And we talk a lot about the impact of AI on businesses on this podcast, but looking towards the future, what emerging technologies or innovations in AI do you find most exciting for their potential to transform healthcare and drug development? What are you seeing here? What excites you? What makes you want to jump out of bed in the morning and work on? Well, you know, I think that there's two pieces of it.

[00:15:55] On the discovery side of things, you know, although, you know, I mentioned that I think that we're a little ways out from realizing the full value there. Admittedly, I'm a huge fan of following the space. Many of the big research labs either have spin outs or, you know, have a section of the company that is focused on drug discovery outcomes.

[00:16:21] A good example there of a spin out is isomorphic from Google. And then, you know, you have recently, for example, OpenAI partnering with Sanofi and Formation Bio.

[00:16:39] So I just like watching the space because just like the broader development of artificial intelligence towards, I guess, a goal for the big labs of artificial general intelligence. There is a race there as well. The belief is that, you know, if you can build a generalized system to to solve this problem, then you unlock a massive amount of value for humanity.

[00:17:06] You know, who doesn't want to follow that and understand the implications? So I'm very excited about what's happening there, although, like I said, I think we're I think a lot of the capabilities are currently overblown. One of the things that I, you know, followed very closely and when I was in graduate school working on computational biophysics was the competition yearly where people had to basically take the sequence of a protein.

[00:17:33] And then they were supposed to predict the three dimensional structure for it. That was an unsolved problem for a long time. The people from DeepMind and a few others recently received a Nobel Prize for solving that problem with alpha fold, which is wonderful. But solving the protein folding problem doesn't necessarily mean that you can automatically cure diseases with it. There are many downstream problems that need to be solved.

[00:18:00] And I think that that's sort of one of the things that, you know, people overlook when thinking about the implications of a drug discovery platform.

[00:18:09] So I'm excited to see when companies either do or at least start talking more broadly about solving that problem more comprehensively and how they're doing it, because that I think will give a bit more credibility to the idea that artificial intelligence is going to solve that problem.

[00:18:33] Beyond that, it's not an area that I'm involved in, but I think the space around AI note taking and things like that just as an accelerant for clinicians. And that space is exciting simply because it'll lift a lot of the administrative burden, you know, from from those tasks, which is, you know, a big part of the job and free them up to do, you know, some of the more challenging, valuable work.

[00:19:03] That impact is sort of the same thing that we're after with our platform. So naturally, I'm excited about that. I'm very excited about the impact that we're making in the quality and regulatory space, because, you know, as far as I'm aware, we're the first ones that are that have a real solution to that problem. And it's just exciting to be at the forefront of it. And BioFi operates in a highly competitive and fast moving sector.

[00:19:32] So on behalf of other people working in the industry, how do you sustain innovation within your team? And what strategies do you employ to keep BioFi at the cutting edge of AI and drug development? Because it's easy to get so bogged down with some of the regulatory and complex processes out there and keep momentum building. But any tips you can advise on how you keep that momentum going? Definitely.

[00:19:58] So it is hard, especially being an early stage company, because you're always focused on delivery. You want to, you know, engage with your partners. You want to deliver on time the best products that you possibly can. And those are usually very well defined.

[00:20:15] What I will say is that I think that there has to be efforts separate from, you know, development and engineering specifically for the product that are based purely around, you could almost say speculative research. You know, obviously it's directionally towards the goal, but there has to be time to do so.

[00:20:34] So the way I like to think about that is kind of separating out either individuals whose tasks are purely research related and have no or minimal responsibilities outside of that so that they can focus specifically on it.

[00:20:52] And, you know, like, for example, one of the things we're doing at the end of the year here is the team has a very large chunk of time where they're going to be focusing specifically on what I call just personal research projects. You know, there's. There's ideas that pop up all the time during the year around things that could create value or better ways to do things, but oftentimes we don't get the opportunity to pursue them right then and there.

[00:21:22] So it's a time and a space for innovation and exploring those ideas and seeing what impact we can create out of those. So short answer is making sure that you either have a separate team that is responsible purely for research and at the minimum for the people who are involved heavily in the day to day delivery, making space for them where they can actually pursue those ideas to is very important.

[00:21:52] And although you work in a very serious industry that can make a massive difference in the world, I suspect that behind the scenes you've picked up more than a few stories along the way. So to to end on a lighthearted moment here, can you share the funniest or or maybe most interesting story that has happened in your career? I'm sure there's a few you can share and a few you can't. But what story would you like to leave everyone listening? Sure. I think this one is funny in hindsight to me.

[00:22:19] And it also, I guess, you know, it helps. It also helps people understand that although on paper it looks like, you know, maybe I've had success after success. It's not always the case. When I was starting to do a lot of the work in artificial intelligence in Johnson and Johnson,

[00:22:38] I had, you know, the ambition to create artificial intelligence driven process optimization for some of the antibody manufacturing that Johnson and Johnson was doing. So I had conceived of this idea. I had done some basic testing on the data and I had to go and present the idea to the entire site that was responsible for creating this particular molecule that I wanted to work on.

[00:23:10] And after I got finished, a person who I didn't know, his name is Tom Merkel, if he ever hears this, stood up and basically said, there's no way that's going to work. I'm never going to let you touch this process. And I basically felt like I got hit by a Mack truck and I wasn't really sure what to say. And he wasn't trying to be combative. He had a massive amount of experience in that product.

[00:23:37] And he knew, you know, every nook and cranny about how it worked and how it was created. And it's just a great example of, you know, the barriers to getting things like this done. So what ended up happening post that is he he was willing to work with me and say, OK, like you think it's going to work? Show me show me exactly what you want to do with this.

[00:24:01] And over time, he became a initially a partner, then an advocate and a great supporter and friend. But it's funny for us to look back at that where we, you know, butted heads in front of effectively the entire site. And then, you know, after we worked on things together and he really saw the value, he became somebody who was, you know, preaching it just as much as I was.

[00:24:27] So it's a fun memory because of, you know, how it turned out going all the way from, you know, something that completely shocked me to the point where I wasn't really sure what to say in front of probably, you know, like 100 or so people to the point where, you know, we're great friends. And it ended up turning out as a massive success. Wow, what a great story. I love stories like that.

[00:24:54] And for anybody listening that just wants to find out more information about anything we talked about today, maybe even continue this conversation, ask a few questions, et cetera, or just look for more details. Where would you like to point everyone listening? Sure. So if anybody's interested in finding a little bit, finding out a little bit more about what we're doing, our website is biofi.ai.

[00:25:18] And then my email specifically is dave at biofi.ai if anybody wants to get in touch. Awesome. Well, thank you so much for sharing your story today. I, for one, have looked hearing about how BioFi is leading this revolution in AI drug development with this platform designed to assess biological feasibility, better predict the likelihood of clinical trial success. That's so great to hear stories about how AI is being used.

[00:25:45] And we're looking beyond the hype now and into real world use cases and making a big difference through technology. So thank you for shining a light on this today. Appreciate the time, Neil. Thank you. So as we wrap up our conversation today, I think it's clear that BioFi is redefining what's possible in drug development. And they're doing that through the power of AI.

[00:26:04] But if we look beyond that buzzword, we're seeing real world implications here from improving clinical trial success rates, real measurable impact there to automating regulatory compliance. Dave and his team are not just innovating. They're almost paving the way for more efficient and a more impactful pharmaceutical industry. But what is it that excites you about the future of AI in healthcare? What concerns do you have? And how do you see this technology addressing challenges in drug development and beyond in the industry?

[00:26:35] Please share with me your thoughts. Let's keep this conversation going. You know where to find me. Techblogwriteroutlook.com, LinkedIn at Neil C. Hughes. But until next time, stay curious. Keep exploring this transformative power of technology that we're seeing impacting every industry right now. And why not join me again tomorrow? We'll have another guest on another topic about how technology is transforming our lives, work and even world. But that's it for today. Speak with you all bright and early tomorrow.

[00:27:05] Bye for now.