Healthcare Revolutionized: AI's Role in Drug Discovery

0:00One of my favorite things to talk about on the podcast is using AI to help improve the medical field to help improve humans lives through science. There's a lot of really cool examples of this right now happening. And one recent report just came out of the University of Edinburgh. And essentially, they used an AI algorithm to help find drugs that could combat aging. And they actually found three different chemical compounds that would be able to help with this. So today on the podcast, we're going to be breaking down what this study found, how they found it, and also what ways we believe that AI is going to impact healthcare, is going to impact drug discovery, and overall human health and well-being.

0:43So, without any further ado, let's dive into it. So, this new research out of the University of Edinburgh essentially found three different chemicals that target faulty cells, which are linked to a range of age-related conditions, and they found this using a method which is hundreds of times cheaper than standard screening methods. according to researchers. So essentially, the findings found that the drugs could safely remove defective cells known as sensent cells. So essentially, those are ones that are linked to conditions including cancer, Alzheimer's disease, and also declines in eyesight and mobility, right?

1:19Typically things that we see as people progress and get older. So things we would, you know, associate with aging. So while previous studies have shown a lot of early promise looking for new aging drugs. Until now, very few chemicals could actually safely be used to eliminate these sensent cells that essentially have been identified, you know, and as causing these effects. And typically, they're called sensolytic drugs. And they are often highly toxic. They're not good for, you know, really, they don't have a good effect on healthy cells in the body, researchers say. And so this led researchers at the University of Edinburgh to look for new drugs and new compounds to try to help achieve the same effect without some of the negative effects and side effects.

2:06So they used machine learning and they were able to use this in a way to find some of these drugs with AI. Essentially, they developed a machine learning model by training it to recognize the key features of chemicals with some of these sensolytic with that essentially had sensolytic activity. And they were using data from more than 2 ,500 chemical structures, which they had pulled from previous studies. So the team then used the model they developed to screen over 4 ,000 chemicals. And they used this to identify around 21 potential drug candidates for experimental testing. Now, this is very similar to a recent report that we saw where essentially they have the same process.

2:47They identify some of the traits they are looking for. They train a model off of that, and then they feed in, you know, thousands of different chemical compounds into that and have it identify some potential drug candidates. And essentially, this process cuts down the time of screening by like an insane amount, right? Like if instead of having to go to a lab and test 4 ,000 different chemicals, you could just go and test 21 potential chemicals. That saves you a ton of time. So that's what they did. uh dr diego aryazun who was commenting on this said this study demonstrates that ai can be incredibly effective in helping us identify new drug candidates particularly at early stages of drug discovery and for diseases with complex biology or few known molecular targets um lab tests in human cells essentially revealed that three of those right identified 21 potential ones they took those to the lab uh they did some lab tests with human cells and they revealed that three of the chemicals, one called ginkatin, one called periplocin, and one called oleandrin, were able to remove sensent cells without actually damaging the healthy cells.

3:56So this is not something that we were previously able to do. You know, we had compounds that were able to do this, but they were damaging the healthy cells as well. So now they're able to find three that did not damage the healthy cells. All three of those, I think this is actually kind of interesting, are natural products found in traditional herbal medicine. That's what the team says. and oriandrin was found to be more effective than the best performing known sensolytic drug of its kind so they literally discovered a new compound I guess discovered is an interesting word because it was used in traditional herbal medicine so someone else perhaps discovered it first but they're able to you know scientifically narrow that down as being you know the main candidate and this was doing a lot better than any of the other drugs that have previously been developed So Dr.

4:41Vanessa Smir Burton, who is from the Institute of Genetics and Cancer School of Information, said, This work was born out of intensive collaboration between data scientists, chemists, and biologists. Harnessing the strengths of this interdisciplinary mix, we're able to build robust models and save screening costs by using only published data for model training. I hope this work will open new opportunities to accelerate the application of this exciting technology. I think something that's really amazing is that they were using published data. So data that already existed and no one had connected the dots to really identify these different chemicals as being the ones that would solve this problem.

5:24So the study published, the study essentially they went on to publish this in the Journal of Nature Communications. It was supported by the Medical Research Council, the Cancer Research UK, United Kingdom Research and Innovation, and the Spanish National Research Council. So all in all, a really incredible breakthrough and discovery. And I really quickly just wanted to bring up the thought experiment right now, where essentially looking at the impacts of AI on healthcare over the next few years and over the next decade, I think is going to be absolutely insane. If you think about it right now, more than 5 billion people around the world have access to a mobile device, a mobile phone.

6:09And now that we're starting to add AI onto all of the tools there, right? Think of, for example, Google Lens can now go and identify a skin condition, right? You can point it at if you have some like eczema on your arm or some sort of growth or a mole. You can take a picture of it with Google Lens and Google Lens will be able to identify what that skin condition is. now imagine um you know you kind of extrapolate that a bit and look at maybe not just in the next five years but maybe even but definitely in the next 10 years think of the fact that this ai technology mixed with mobile devices is going to completely democratize healthcare for a vast majority of the world especially people in underdeveloped areas imagine if for example you had a sore throat and instead of having to go to a doctor you could take your phone turn your you you know, turn your flashlight on, shine it down your throat, it could just take a look at that, take a picture of that, and it could go diagnose you with what you need.

7:03You know, maybe if we want to have some sort of verification, it can take a picture of your throat and send that over to someone to, you know, a pharmacist or someone to automatically start pulling, filling out your prescription, or any other, you know, physically visible, obviously, this isn't going to work with something that requires blood samples, but any physically visible ailment that you might have, it could look at and diagnose very quickly and very effectively. So this is one area that I am super excited about and super excited to continue even to follow how AI is going to revolutionize healthcare, really democratize it and help an incredible amount of people able to have access to good healthcare.