Broadcast Retirement Network’s Jeffrey Snyder discusses the use of Artificial Intelligence in the creation of new drug with Early-Stage Biotech Scientist Andrew R. Snyder, PhD.
Jeffrey Snyder, Broadcast Retirement Network
Joining me now is Dr. Andrew Snyder. He’s a drug researcher and developer and scientist, and for full disclosure, he’s also my brother. Dr. Snyder, Andy, good to see you this morning.
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Nice to be back with you, Jeff.
Jeffrey Snyder, Broadcast Retirement Network
Yeah, it’s always a pleasure talking with you. Before we get into artificial intelligence and the drug development process, I want to check in on the investment and innovation. How’s that going?
Are you seeing from your relationships in the field, are you seeing more investment now than versus last time you were on the program?
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Maybe incrementally. I think there are some small cracks. Last time we spoke about IPO windows, maybe that’s cracking open a little bit.
There’s certainly a lot of big news on the M&A front, and all of that will trickle down into the innovation landscape. I think we are seeing positive signs in our environment.
Jeffrey Snyder, Broadcast Retirement Network
That’s great. That’s great news. I think it’s great for patients.
It’s great for all of us because we need these innovations to continue to live longer and happier and healthier lives. All right, Andy, let’s talk first about the traditional drug development process. It takes a long time.
Can you just give us the broad brushstrokes of what that looks like from start to finish?
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Yeah, that’s a fantastic place to start, Jeff. My experience is generally on the novel side, so new targets, maybe an existing modality, but something new. You’re going to just say, these are rough numbers, but this is going to give you a timescale here.
You’re going to say it costs $2 billion to bring a drug to market, and you’re going to say that it takes 10 to 15 years. These are the costs associated with bringing a new modality forward. There’s lots of bottlenecks along the way.
One brute is taking a novel academic target and understanding if it is a good target or not, screening that target for maybe a small molecule that modulates it, and then developing that further preclinically to work in animal models. That’s the preclinical phase, and that’s where I spend all of my time, but that’s really where the minority of the investment takes place. It’s then when you enter the clinic where you have high-value, high-dollar studies in patients, sometimes for the first time with a new modality or with a new target.
Jeffrey Snyder, Broadcast Retirement Network
Those would, I guess, we would refer to as clinical trials. You’re trying out the drug to see it’s a process, and you’re trying it out to ensure that it actually works and does what it needs to do and if there are changes along the way. Now, let’s talk about artificial intelligence because you read the same things I do, but you see the news reports.
AI has just been infused everywhere from manufacturing to financial services. A lot of people use JETGPT. We don’t on this network, but how can artificial intelligence help eliminate some of those bottlenecks you were talking about?
Maybe, could it actually help speed up this 10- to 15-year process you were talking about?
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Jeff, that is the hope. We’re seeing AI permeate all phases of the drug discovery process. Let me give you some pointed examples about what I’m seeing as I read the news, as I talk to people in my field, as I look to the future.
I think, you know, let’s first talk at the discovery side. Okay. So, there are tools called AlphaFold, which help you predict protein structure.
And what this might have the effect of doing is give chemists and other drug developers a better and faster chance at finding the right molecule now that they’re able to have a 3D structure in hand. Similarly, there are even just these monumental approaches to have autonomous laboratories that are not only working on their own, you know, moving plates around the lab, but they’re, quote-unquote, thinking on their own and testing their own hypotheses, right? And so, there’s companies like Lila Biosciences here in the Boston area that are really pioneering this approach.
I think I was just looking, you know, they call it laboratory superintelligence. And they’re really trying to spur on innovation. And just let me give you a third example what’s happening on the really early target side.
There’s a company called Tahoe out in California, which is working on these single-cell perturbation datasets. They’re treating cells and they’re trying to understand the interactions between different molecules and different proteins. And they end up making a lot of this data publicly available, but they’re also using it to make key insights into drug targets, right?
So, they’re trying to find connections between the data that will maybe give a better target than they would ordinarily have found.
Jeffrey Snyder, Broadcast Retirement Network
So, you know, you have typically done, I would say, I’m going to call it the trench work that’s testing. Does the future Dr. Andrew Snyder, are you part data scientist, part researcher? So, you’re actually shifting.
Someone in your field is actually kind of embracing this technology and shifting forward of their approach. Because I would imagine with these datasets and this new technology, maybe you don’t have to do some of the same tactics. Maybe you adjust your tactics.
Again, I’m a lay person here, but what I’m thinking about is what is the future research scientist, someone like yourself, look like? How do you evolve into that next, using this technology?
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Well, look, I think some of this is kind of TBD. Let’s see some of this play out. I have people on the chemistry side tell me that some of this needs to be empirically determined, right?
The kind of way that we do things might continue to be the way that we do things. But there are other datasets that can certainly be leveraged. And I think it’s understanding where those bottlenecks are and where we put those tools into place, right?
Especially companies, you know, like large pharma companies that have massive, massive, both pre-clinical and clinical datasets. AI is going to be really key. How can we mine that data to make connections that would have taken us forever to make?
Or maybe we didn’t see them. And so, you know, whereas, you know, I don’t know that I would be the data scientist, but I think it’s got to be on everyone’s radar as we move forward through all of these different drug discovery and development processes.
Jeffrey Snyder, Broadcast Retirement Network
Yeah, at least connected to your hip, so to speak, so that you’re kind of, you know, I kind of liken it to, we’re both Orioles fans, unfortunately, in some regard. The sports now, they have a lot of statistical people that work directly with the coaches. Maybe it’s kind of that analogy.
And then I guess my last question, and we’ll bring you back next month, of course, I want to talk about the regulatory process. Because if you’re using artificial intelligence, I would imagine that the Food and Drug Administration, other regulatory entities would need to take into account that some of these data models are being used in order to create new drugs. I mean, I think that kind of changes the traditional approach to the regulatory environment.
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Well, I don’t know, they’re still going to require the same data packages, you know, the drug has to be safe, and it has to be efficacious for it to get approved. And of course, there are certain situations with high amounts of unmet need, where there are, you know, accelerated approvals, etc, based on historical data sets. But I think the data package is going to remain the same.
You know, I just think it’s a case where we have the opportunity to accelerate drug discovery and development. You know, Jeff, there’s one other place, you know, I was reading somewhere, and I don’t want to, you know, sidetrack the conversation, something like 10% of phase one assets end up going to approval. And so 90% of them don’t.
For a variety of reasons. But if we can use AI, and this is being done, you know, throughout the ecosystem, to target the right modality at the right time to the right patients, maybe we draw back some of those failures, right? And maybe we can target these therapies to the right people.
And so, you know, we’re seeing this throughout the entire lifecycle here, from beginning to end, I think, you know, we’ll see what the future holds. But there’s a lot of promise that it can impact multiple steps of the drug discovery process.
Jeffrey Snyder, Broadcast Retirement Network
Yeah, we didn’t even get into the experience versus non experience. I mean, you have decades of experience of building drugs, and processes. That’s important to partner with the data scientists with the with the data sets that AI is looking at.
Dr. Snyder, Andy, we’re gonna have to leave it there. It’s always great to see it. I look I wish I could see in person a little bit more.
Maybe we’ll see each other very soon. But we look forward to having you back on the program again, very soon.
Andrew R. Snyder, PhD., Early-Stage Biotech Scientist
Thanks, Jeff.