Invest Like the Best with Patrick O'Shaughnessy
· Patrick O'Shaughnessy
[00:00:00] S0: What about imaging? Both skin imaging, interesting companies doing that, like Neko. Deep body imaging, companies like Prenuvo. It's quite expensive, but as I think about the health stack as something that we could do that's preventative, how do you feel about those technologies which are relatively new but do try to take a picture worth a thousand words inside your body and identify a tumor early or something like that? Do you think that's worth it?
· Alex Karnal
[00:00:00] S1: To me, more data is worth it so long as you can put it in the right context. When we get the result of a colorectal cancer screening test, we know the context for that and we actually know whether we have something that's useful or not. The pictures that we can get from groups like Prenuvo and others are powerful because we can see progression over time and we can use that data, and particularly as AI is going to increasingly become our medical home. Arming our AI system with that information will be super helpful. The thing that I worry about, though, is as you measure more and more different dimensions, with each of them having a material false positive rate, the odds that you come out of that experience with a false positive that can be disruptive in life becomes material. It's not to discourage anybody from doing these. It's just to make sure that any data that we're collecting, make sure we put it into the right context set so that we can have a framework for how we deal with whatever information comes from taking that test. So for me personally, I'm not yet pursuing those. I'm very much focused on areas where if we're going to go and do a test, it's going to have a high probability of capturing what I'm testing for, and a low probability of giving me a false reading. And I think the more information that we can accumulate with really high fidelity, the better we'll be at being proactive about our health.
· Patrick O'Shaughnessy, Alex Karnal
[00:00:00] S0: Everything we talked about today was discovered and developed by incredibly talented people, both scientists – people in the medical field – but then business builders that turn these things into great products and great companies. Maybe just describe so as not to take it for granted what that process is in the first place. Back to biology is complicated, it is un-freaking-believable we can do the things you've been describing so far in this conversation. [00:00:00] S1: It's pretty wild, yeah. [00:00:00] S0: And we're on a trajectory where the granularity with which we can make edits, for lack of a better term, into what's going on in biology is probably just going to continue to accelerate. So I want to understand that process itself and then we can talk about how AI figures into that discovery process.
· Alex Karnal
[00:00:00] S1: First of all, what are the two key ingredients for discovery to happen? You need scientific instinct married with tremendous amounts of capital. And I think you need both of those to converge in order for there to be a breakthrough that allows us to live healthier and longer lives. Let's break those into their pieces – the science of medicine and the business of medicine, because I think they're both independently interesting. Oftentimes what ends up happening to discover a medicine is a scientist, for whatever reason in their journey, gets inspired to spend their lives trying to crack the code in a certain disease. What tends to happen is you've identified a problem that you want to solve – just like in any consumer or tech product, what's the problem we're trying to solve? You then have to go through a tremendous amount of research to understand every single thing you can about that problem. And most importantly, what could be driving it? What's developed over time is a method, probably one of the most important methods in the history of humanity, called the scientific method, which is the systematic way by which a scientist moves from focusing on a problem they want to solve to actually solving it. And there's a couple of discrete steps in that scientific method and tremendous amount of iteration that leads to cracking that code. Once you've identified the problem that you want to solve, the first step to solving that is come up with a hypothesis for what could be driving that disease. And the way scientists tend to do that is – everything from the PCSK9 example, where scientists are observing nature. "Where are there populations of people that have an advantage? Okay, let's go figure that out. That's cool. Let's go understand it." All the way through to combing through thousands and thousands of pages of the literature to understand what people have tried to crack a code that still exists today that you want to go crack. And that's a deep literature review. Fortunately, there are tons of papers that exist on successes or tiny steps to successes that exist and are available for anybody. The unfortunate part of the literature is that what's often missing is all the failures, which would be even more informative than the successes. So generally speaking, when a scientist wants to go solve that problem, they're going to observe a population.
· Alex Karnal
They're going to start reading a ton of research. They're probably going to put both of those insights together and use that foundation to form their first hypothesis for what could be the solution to the problem. So from having a hypothesis, then the amazing scientists can then go figure out what experiments do I need to run to test that hypothesis? And even more, what's my prediction for what should happen in those experiments? And those are really important steps because the design of the experiment and the quality of the data that comes out from that experiment will either confirm your hypothesis and you move forward, or it will reject your hypothesis and then you go back through and you loop to your next hypothesis and your next set of experiments, and you continue to loop and loop and loop. So what scientific method generally looks like is hypothesis, experimental design. Execution of that experimental design. Review of the data that comes from that experiment reading out. Test that data result versus your original hypothesis. Either confirm the hypothesis or re-loop, and re-loop and re-loop and re-loop. And so there's this iterative process that tends to begin its journey from the literature, very quickly moves into what we call in vitro experiments or experiments that are done in petri dishes. And as we start to develop evidence and support for a hypothesis in cell-based experiments, it tends to move on to small animals and then larger animals. And then it gets to the point where the totality of the data and the evidence that you've collected from a hypothesis perspective to the cell-based experiments, to the animal-based experiments – you put that all together in a beautiful package that you present to regulators and you say, I think I've cracked the code. I think I can move us forward against disease A, B, or C. And here's the hypothesis that we had. Here's how we tested it. Here's all of the data we've generated that helps me, the scientific team, to believe that we might have cracked this code. And now we would like your approval to go start and agree on what studies should look like to test how toxic or how dangerous this breakthrough could be for somebody.
· Alex Karnal
And if we can show that there's good margin between really high doses and where that toxicity starts to come in, and the doses that we've proven from our experiments are likely to help somebody live a longer, better, healthier life – we want your permission to then go and actually test that in a person for the first time. And this is incredible. And that process that I just described, you can imagine most of them fail. But for the ones that do succeed and get the chance to then go into toxicity studies and ultimately into humans, those are the ones that get really exciting. But they take years and years of iterating and looping, to go from that beginning idea, through literature, ultimately to the animal study that makes you believe you've actually cracked that code. And then, of course, from there – this is so much more of a marathon than a sprint, because you've spent years trying to crack the code, and now that you have, guess what? Now you've got another five to seven years worth of clinical studies you have to do, to prove that in people, the idea you had for medicine actually has the effect that it's intended to. And that, per our discussion earlier, that benefit-to-risk equation is favorable to the point that regulators should approve it and let it come to the market.