Tal Zaks
speaker
203 appearances
1 recordings
1 series
first heard Jan 2025
last heard Jan 2025
Tal Zaks’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
And some of it has to do with the nature of the beast, if you will. It's not a system we designed. So we're constantly learning about it versus engineers. But some of it, and this is one of my biggest lessons from Moderna, it's also a different mindset. Engineers and physicians are almost diametrically opposite in how they think.
And I had the good fortune in my life to work with, I think, one of the most brilliant engineer leaders, Stéphane Boncel, the CEO of Moderna. And what it taught me is that engineers have two things that physicians struggle with. One is vision and the other is process. So what do I mean by vision?
Well, if you look around our life, take a look out your window if you live in the city, everything you see is a function of vision of people who came before you. it would not otherwise have been there. That's why we love nature so much, because it's uncluttered by visions of other people. But if you live in any other environment, you're constantly faced with a vision.
So you have to ask yourself, well, what's my vision for the thing I will leave for the future? And then to enact on it, okay, it requires a process. So what does it actually take to get there? And here's an interesting thing. If you ask an engineer an engineering problem in their domain, Say, I want to get to the MOOC. Is it possible?
Well, anybody who knows the domain will tell you, yeah, it's possible. Here's what it's going to take. And they can even map out the resource and the likelihood of it getting there. Well, you ask a physician, hey, I want to cure cancer. Is it possible? And the answer is, I don't know. I mean, I can think of some approaches, but we got to go try them.
One of the first things I did when I joined Moderna, and probably the scientific thing I'm most proud of in my time there, is actually not the COVID vaccine. It's something called a personalized cancer vaccine. So we've learned enough about cancer to know that everybody has a different cancer. Everybody's immune system is different.
And we know that mRNA is a phenomenally good platform for making vaccines. So can we immunize people against their cancer? Now, to do that, we'd have to figure out what each individual needs to get immunized against. And so it would be personalized. And that's a very challenging, complicated task. It's expensive. So we set out to do that.
And as we were starting down the path, I remember we had a phase two study that was running. So that's basically a study where you give half the people your vaccine and half the standard of care. And the question was, are we actually going to be able to prevent some cancers from returning? relative to the computer.
And Stefan used to stop me in the hallway and said, Tal, is this thing going to work? And I'm like, Stefan, I honestly don't know. What I know is I feel I've been put on earth to do the experiment. And we've been fortunate to align the technology and the resources, including the financial ones, to run the experiment. but I'm not going to predict whether the experiment's going to work.
People have been trying to do cancer vaccines. I've been trying to do cancer vaccines since I was a postdoc, and so far we haven't. I think this one has a chance of working, and here's why. But until we test it in the clinic, we won't know. And we've been fortunate that that phase two actually read out with quite a successful readout last year.
In fact, the phase three now is enrolling, and hopefully we'll have a personalized cancer vaccine on the market before too soon.
So that's still in the future. The way this works is people with early stage cancer, specifically, we started with skin cancer. So cancer has been diagnosed, but it's early. Likely a surgeon will cut it out. And then there's some probability that it will come back. And what can we do to improve the odds that it won't come back?
In that context, at least as far as a randomized phase two goes, when you give a personalized vaccine to those patients, it cuts the recurrence rate of cancer by about half, by about 50%, which is quite significant.
So I'd break it down to three phases. And I do think that the world is improving. The first phase is, is this target for intervention actually relevant to the disease? That's basic biology. And I think we're making great strides there in understanding biological processes. Some of it is big data. Some of it is just old school grunt work.
But there's a lot of tools that have been developed that are being deployed that are making this much more accessible. to us understanding disease better. The second has to do with what's called drug discovery. Okay, so I've got a protein whose function I want to alter. What is the ability to actually discover a new chemical entity or a new protein entity or nucleic acid entity that will actually
interfere there, that will actually do the pharmacological effect. And there, I think you're seeing a very significant deployment of these modern AI tools across the industry now.
And it's very quickly becoming, to a certain degree, commoditized with the advancement of alpha fold predicting protein structures and people applying the same kind of tools into chemical discovery space to come up with new chemistries.
I've seen people even apply these tools into figuring out these lipid nanoparticles that will shepherd mRNA into different tissues to come up with better formulations and ways of bringing that medicine into the right places in the body. So I think drug discovery is getting a leg up and a significant one from the various applications of AI tools.
The part where we're still behind is in what's called development or clinical development, putting it in people. There's no shortcut here. We don't have a holistic model of a human being. We have made progress in understanding what natural outcomes are for people with high quality, big data sets that you can apply machine learning tools.
So our ability to predict outcomes on the control arm is getting better. And so people are leveraging those tools to make trials shorter and smaller. But in the end, the gold standard is and will remain, okay, I have to put it in people and see how those people respond and make sure that the people who get this have a better outcome than the people who don't.
Showing 21–40 of 203 · page 2 of 11
← Previous
Next →