The Chris Voss Show - The Chris Voss Show Podcast – Dr. Lana Feng, CEO and Co-Founder of Huma.AI Using AI To Speed Up Drug Development Time, Lower Costs and Pricing

Episode Date: August 2, 2023

Dr. Lana Feng, CEO and Co-Founder of Huma.AI Using AI To Speed Up Drug Development Time, Lower Costs and Pricing Huma.ai Huma.AI Begins Where ChatGPT Ends - Leading Generative AI Platform for Life Sciences - Delivers results that are always up to date. Outputs content with zero character or word limits. - Analyzes data with complete privacy - Works across all data silos - internal and external - Provides references to internal and external data in every result - Applies to a wide variety of Life Science use cases - Validated at 97% accuracy in current implementations

You wanted the best. You've got the best podcast, the hottest podcast in the world. The Chris Voss Show, the preeminent podcast with guests so smart you may experience serious brain bleed. The CEOs, authors, thought leaders, visionaries, and motivators. Get ready, get ready, strap yourself in. Keep your hands, arms, and legs inside the vehicle at all times because you're about to go on a monster education roller coaster with your brain. Now, here's your host, Chris Voss. Hi, folks. This is Voss here from the chrisvossshow.com, the chrisvossshow.com. Welcome to the big show. My family and friends. For 14 years going on 1,500 episodes, The Chris Fosh Show has been coming to you,

broadcast to wherever you are, in your car, in your house, in your home, your work, whatever. I mean, you could be living on a Viaduct, and you can listen to The Chris Fosh Show, and it's free, too. So that's probably why you're living on the Viaduct, listening to The Chris Fosh Show, because, I don't know, it beats just living on the Viaduct.

Who's doing that? Everybody lives down by the river these days anyway. So welcome to the show, my family and friends. We're going to be talking about AI. And we're going to make the show really intelligence, but with artificialness. I don't know. See what I did there? I don't know.

It sounded funny in my head at the time. And we have an amazing guest, CEO, who's going to be on the show to talk to us about it. But in the meantime, as always, give us five stars over there on the iTunes. We certainly appreciate those referrals. And also go to Goodreads.com for just Christmas. YouTube.com for just Christmas. The big LinkedIn newsletter and LinkedIn.com for just Christmas. We're going to be talking with Dr. Lana Feng today.

She is a CEO and co-founder and has over 20 years of experience in biotech pharmaceuticals and focused on precision medicine. We're going to be talking about her company, Huma AI. And she worked with Novartis Oncology Business Unit and built the biopharma division at GenOptics, I think, increasing sales from zero to 45 million in five years. She also forged alliances with pharma companies and provided CDX development for targeted therapies. Welcome to the show, Lana.

How are you? Great, great. Glad to be here. And it's wonderful to have you as well. Give us your.coms or your.ai, maybe I should say, so people can find out more about you guys on the internet. Yes, it is www.humai.ai. And I'm on LinkedIn as Lana Fan.

There you go. So tell us about this company. Give us a 30,000 overview of what you guys do there. Okay. So what we do is to automate a lot of the data analysis and we can use AI to accelerate the development of life-saving drugs, bring them to market faster. There you go. And I know it takes a long time to develop different drugs and pharmaceuticals. Is that correct? That is correct. And part of the reason that it takes 10 years and $2.6 billion to develop a drug, that is

why they're so expensive, is because of this bottleneck on data. So this is a problem that extends into the healthcare as a whole, healthcare life sciences, because it's a highly regulated industry and it's highly complex. So we have a deluge of data. There's a massive amount of data. It could be electronic health records, could be your payment data, your insurance data,

from our clinical trial data all the way to scientific publications, right? So there's massive amount of data. And these data, we call them in sort of different disparate systems that don't talk to each other. So this is another uniqueness for our system. And then lastly, is that 80% of that data was what we called unstructured, meaning they're not numbers. They are, you know, free text, right? They are documents. So the industry relies on really kind of experts like myself reading those documents in order to come up with intelligence insights. So we're really, yes, go ahead, Chris.

I was just saying, wow, what sparked the creation of this company and what made you want to do it? So maybe I can start with my kind of personal journey. Yes, please. So I was, I grew up in a family of physicians, right? My parents, so wanted me to go to med school. I was, but, you know, being the silly, you know, ambitious young myself. And I said, you know, I wanted to get a PhD. I want to go into drug development so I can treat patients at scale. So that kind of became my journey. Instead of medical school, I went to graduate school and got a degree in genetics,

and then did a postdoctoral research at UCSD in San Diego in California, and then directly went into industry. So this 20 years of experience in precision medicine is very much focusing on developing cancer drugs. You know, precision medicine is about giving the right drug to the right patients at the right time. So when I started out, it was kind of really at the forefront, right? What do you mean when I'm diagnosed with lung cancer, I have to submit my sample for genome sequencing, right? But then now it's common practice. So really kind of very, very exciting journey. So you mentioned about genomics. So I was hired very always kind of entrepreneur. So I was an entrepreneur, we call it,

basically building new businesses within a larger entity. So at Genoptics, I was hired to build a biopharma business, right? Offering the capabilities in cancer diagnostics and testing to clinical trials so like you said they grew that very rapidly and then Novartis came in and bought the whole company wow or for so that's kind of my journey from was kind of a small to mid-sized startups to a global global massive uh pharma company so that's pretty awesome yeah yeah so the clinical trials we were doing was like 30 50 trials all of a sudden you have like 300 trials right well speaking of like doing stuff at scale and as a um as a kind of i call myself domain expert

right so i you know design clinical trials and what have you it became kind of not scalable so manual curation of all these data right so can we do this a better way right and rely on data scientists and they typically they're not we call domain experts right they typically came from tech they don't understand the problem that we solve. For example, I think I'm kind of going into the weeds, but just to help the audience understand, right? Precision medicine, we're enrolling patient with a particular profile. For example, a genetic mutation, right? If this genetic mutation has 10% prevalence,

meaning 10% of lung cancer patients have this mutation. And that means I need to screen 1,000 lung cancer patients to get 100. Are you following me? The 10%. So you can get the data, yeah.

You can get enough patients to run trials, right? Google trials typically fail because we can't get enough patients. Oh, really? Yes, that is a huge problem we're still wrestling with. But then if you get that data wrong,

instead of 10%, the actual prevalence, right, the commonality of that mutation is only 1%. So that's a big spread. That means we have to screen 10,000 patients to get to that 100. Holy crap.

Yeah. So there's really those numbers and like it has a serious impact. So that's why we started Huma AI is that can we democratize this whole process using AI? Yes. There you go. And it's built on your website as, let me pull this up here, Huma AI begins where ChatGPT ends. Tell us what that kind of means or refers to then. Great. So for those of you who haven't used ChatGPT, right,

so this is basically the consumer version of, we call them generative AI from OpenAI. So generative AI means that rather than surface whatever is in the, for example, a piece of news or an art or what have you, or book, generative AI means that actually AI is generating new content or new data. So ChatTPT, OpenAI launched ChatTPT in November of last year, so about seven, eight months ago, and then basically went and hit a million users in five days. Oh, yeah. Yeah. So it was a it's I think now there's probably 1 billion users on the platform. So that's generative AI. So you can ask any questions. You can plan your vacation.

You can do all these. You can ask it to write songs. So those are kind of generated content. So what we do is that take those capabilities behind chat GPT. We call them large language models. So GPT models, for example, GPT-4, GPT-3.5, you're going to be hiding under a rock if you haven't heard of this, right?

You are. You really are under a rock. Yeah, so prevalent. You can't stay away from this. So you take them and say, can we make it um you know more private and more secure right to work with our clients enterprise data i like i said we help companies bring life-saving drugs to market so we focus on that life sciences pharma medical device companies so those data are precious is also highly private. Can we make that private and secure? That's the first thing.

There you go. And second. Probably have HIPAA rules you have to deal with too. Absolutely. Right. And GDPR from European data. And secondly, and everyone's probably has heard of AI hallucination, right? Just make up stuff with conviction. Yeah. I think there was a, I think there was an attorney who got in trouble because he used,

he used a chat GPT for his, a complaint or argument. And it just, it just pulled fake cases and he got in trouble. So yeah, you gotta be careful. Yeah, exactly. So,

so what we do then we really kind of use this. We believe, right. Generative AI is actually a very much a co-pilot concept. It has to have this human in the loop. And you have to have experts, for example, look at the generated content and review it and then feed it back into your AI engine. We call it reinforced learning through human feedback. That's how you get accurate. So accuracy is really, really important, particularly in healthcare life sciences, right? We're talking about patient

lives here. We're talking about accuracy. We're talking about medicine. So lastly is transparency, right? We provide citations for the generated content. So it's basically saying, what are my source document actually support this? So why is that important? That means our end users who are typically MDs and PhDs, researchers, what have you, being able to trace it back and seeing is this important?

Is this correct? Can we verify this? So this is very much that responsible AI, the transparency, the accuracy, the privacy that we can provide. There you go. So do you think this is going to shorten the timeline from that 10 years or so to develop a drug? Can this squeeze off a few years? That is such a great question, Chris.

That's why that big brain behind 1,500 episodes. You're an expert in everything i i i try to be i try to be me i'm probably like an amateur and everything but i know i you know i try and put my fingers on all the pies just don't know enough to be dangerous right exactly yeah right exactly that's that's the purpose right yeah it's like can we unshave even just a year off um of this um this cycle right 10 years and 2.6 billion so um typically the industry standard is saying a million dollars a day that's the cost of developing new drug you can you can imagine if we just basically accelerated by a month that's how much savings, right? And then the upside.

Ultimately, this should bring down drug price. Really? That would be awesome. I mean, that would be awesome if we could do it. I imagine, yeah, if it doesn't take as long to develop and the high cost isn't there, saving money can really do a lot. What other benefits to life science industry do you think will come

from generative ai oh it's so massive right if you look at the uh sort of um the development cycle you start from drug discovery right and that's actually the original playground for generative ai right to sort of using molecule models to, can we create new molecules that could become medicine? So, and then once that is discovered, you go through kind of animal testing and you apply to the FDA to start clinical trials, putting this new medicine into patients, right? Into human bodies. And then that's where we hear, you know, clinical trials and phase one, two, three. Phase one typically is very much just finding, is this drug toxic? Finding the right dose. And second, phase two, typically you started to say, okay, this drug is safe in patients. Can we see efficacy? Does this actually work to cure disease?

And then phase three, what we call the regulatory approval study. It's a pivotal trial where you submit that data to the FDA for approval. And then post-approval, there are a lot of things happening as well. For example, you have to do post-market surveillance. You have to do post-market vigilance to make sure you put this drug on the market. It doesn't kill patients it doesn't have a lot of adverse events does not have toxicity right that's always important yes very very important right no one wants to die from a bad exactly yeah you also don't want the the drug medicine got pulled off the market right yeah that's true fda there's no yeah exactly it's

you know risk is is the first and foremost thing. There you go. And I think, I imagine, I mean, you told me one of the other benefits might be, I know when, you know, hopefully we don't have another one of these pandemics very soon. But I know, you know, we're kind of lucky that the mRNA vaccine was kind of being worked on at the time but if we get in all these pandemics it might be really helpful to you know speed the speed the approval process through for what was going on with like covid and stuff that is such a good point right i i like to to say that this is actually it's like showed us what can be right what could be it's basically you can bring a drug to market in just several months several years right rather than the 10 year, you could do it in three to four years, right? It's just like, you know, what would be the art of the possible. There you go. And I know there's a big thing, I think it was Biden who approved it

years ago, it was a cancer central database commission, and it was supposed to take all the nationwide, you know, silos of information on cancer and how to treat it and stuff, and it was supposed to put it into one big database. I don't know how that's ever turned out, but something like that that could be cancer, you know, my sister suffers from MS, you know, different things that could speed up any sort of, you know, Alzheimer's is going to become a big thing, of, you know, Alzheimer's is becoming a big thing, dementia, you know, as our aging baby boomer population is, you know, moving into their retirement years, they're dealing with all sorts of stuff and lots of innovations for that. And I imagine that can be really helpful as well. That is really, really helpful. That is what

we call the real world data, right? So this is not data like any companies created, but this is the data coming from the real world. So on the precision medicine side, right, as I mentioned, it started out with cancer. Now it's moving into more chronic diseases, right, such as you mentioned, you know, MS and cardiovascular disease, and this is moving into rare diseases, right, so that data becomes even more important. And aggregating all these different data together, right, you have, you know, your electronic health record, like you said, for example, even maybe encouraging patients to volunteer their data into some kind of disease registries and what have you. So the biggest problem right now for the real world data, right?

Everything under the sun, as I mentioned, is fragmentation. It's very fragmented, right? You have maybe one hospital system offering their data, partner with a tech company to do this, right? You have maybe another hospital system. So this is very much, so you never really have the 360 view of the entire thing. So that's why this government, right? These initiatives are really important.

I think NIH has a million lives initiative to try to really kind of take the 360 view of all the data and even maybe a smaller patient set but then having that government um um initiative it's going to be really really important for our entire um healthcare life science industry yeah and if they i can collect all this data and put it in one place and you know assimilate it in a way that you know we could find maybe the best the best methods and the best ways of processing. I mean, that, that seems like it can make all the difference, you know? I mean, you're trying to, you know, there's so much data that's shotgun out there, if you will, and, and putting it all together in one piece, you know, we can, we can maybe get closer to solutions

and finding stuff. This is really, this, this is really positive because, you know, some people talk negatively some people talk negatively about ai you know uh you know the cue the term in your music here uh you know some people like i'm gonna be on a job because of ai and so it's really good to hear that there are things that are really going to help us and of course we're going gonna go through a sea change you know humanity does this or you go through a sea change. Humanity does this, or you go through a sea change with new technologies. And so it's great that something like this is going on to improve our lives. I totally agree, right? I like to say that AI is not going to replace you. It's the people who use AI will replace you.

Oh, wow. Note to self, get back on GPT today and start working over there. But yeah, I've heard that sentiment. And it's true. It's one of those things that you've got to adapt to and adopt to. It's interesting just what a rage it's become. And I put RPR or sales or different different ad copy that we do in there and holy crap yeah i mean you have to go back in and edit it and put in the human part of

it but uh it's it's just astounding what it does like it's rewritten stuff that i've struggled with to get done just the way i would like it and and they'll do it. And I'll just be like, wow, holy, that took like five seconds, you know? Yeah. Yeah.

So that's, that's really, really it, right? Is that first pass, right? AI does that first pass. And then you,

you as humans, right? We come in or experts in what the topic that we were, we're, we're like, you know, consulting on chat TPT to come in and kind of bring that human factor, right? So I imagine, are you seeing more hospitals or more people in the medical community doing adoption of AI and trying to, is it an industry-wide thing that they're trying to embrace it? So this is different, right?

Than kind of some other technology yeah i i like to say this is probably similar to the the dawn of like the internet right so we're seeing um um there there's there's a method to this madness my train of thought so um so you know when internet um just started right so doctors you know they were like oh you know this is the thing is not good it's it's like you know i'm more accurate right and then as patient as consumer myself i would like go on the internet you know search like oh i have a sore throat what could it be and then bring that print out and it to my doctor will just irritate the heck out of him right right? WebMD says I have everything.

Exactly. I've been on that website. It's like I put in some data and they're like, you've been dead for five years. And I'm like, wow. Exactly. And we always assume the worst, right? Yeah. And we're seeing this as a similar thing, right?

So for example, this is a consumerization of technology, right? We're seeing actually our clients, they've used chat TPT at home, right? Helping their kids with their homework, maybe doing something for their personal use. And then they come into work and saying, hey, why can't we use chat? Or can't we use this for our work, for our data? So we're seeing that with physicians as well, because the barrier to entry is so low in terms of experience, in terms of getting that really like aha insights.

So we're seeing kind of that lower, the sort of the adoption, if you will. There you go. So who are your clients, people out there listening? Who are the people that you're interested in working with that, you know, et cetera, et cetera? So we work with a number of global pharma companies, and we also work with a midsize biotech company. So that's on the drug developer side, right? Bringing medicine to market. We also work with medical device and diagnostic companies, meaning they're designing like, for example, a catheter or some kind of pacemaker or maybe developing COVID tests, right?

So we work with those companies as well. There you go. And so they can reach out to you. I see on the website, they can book a demo if they like and talk to you guys about how to utilize your service for them. That's great. Yes. So we're very much the leader in using generative AI for this life science vertical. In fact, we got a huge recognition from Gartner, as you know, like they are the sort of a technology thought leader, if you will. And then their flagship products are Magic Quadrant, Hype Cycles, and a cool vendor. So we got on 17 Hype Cycles for generative AI. So it's really awesome to be mentioned in the same breath as like Microsoft, OpenAI, AWS.

So that's a really great external validation. There you go. Well, hopefully it can help people lower pharma costs, get drugs to the things faster because, you know, we need them. loved one or someone who's suffering from some sort of disease that is maybe slowly killing them like Alzheimer's or something like that, anything that can be more speedily readily brought to market is definitely welcome, especially if it's safe. So there you go. Anything more you want to talk to us about your company and what you guys do? That is our mission, to really democratize this and maybe eventually bring down the drug price. Another thing I want to close is about this concept of responsible AI and AI for good. I think there are a lot of negative sentiment out there.

I think it's about knowing the risk and really mitigate the risk. There you go. So we're part of some of the consortium, really as an industry, trying to lower the risk and really kind of putting guardrails in place and then utilize AI for good. There you go. And that's important. I mean, I know that there's a lot of discussions about AI and sometimes being used for bad. I know the dark web's up to some stuff with AI. They have their own version of chat GPT that has no morals, ethics, or whatever. But welcome to the web, the internet.

But no, it's good that you guys have that. And I think this is awesome because, like I said, I think everyone has a loved one probably at this point that's suffering from some sort of disease or issue or cancer etc etc and boy if i man if we could figure out a way to beat these things and and live longer as humans and have a more healthier life and we probably need like a chat gpt that calls you every day you know anytime you pull into mcdonald's and says hey don't honor the big mac get out of there. Get out of there. Your heart's whatever. I don't know.

Something like that. That would be – more people need that. Yeah. Don't go with the bad food. Go get some broccoli. Eat that. Exactly, right? The agent, right?

So we're having like now generative AI can provide you with that intelligence and what have you, right? Get that content. But then the next thing really is that we call them agents. Being able to actually make recommendations and help you do stuff. Yeah. It tracks your location. It gives you a little zap when you pull into that fast food thing at McDonald's and stuff.

I'm going to get sued by McDonald's or something after this. Changing human behavior is really hard, right? Yeah, it zaps you when you go down the ice cream aisle at the store. And it's like, get back over to produce. There you go. So there you go. Well, this has been wonderful and insightful to have you on, Lana. Give us the.ai so people can find you on the interweb.

Great. So our website is www.huma.ai. Huma is H-U-M-A. Stands for humans plus machines, where humans always come first. So it's the first letter of human, first letter of machine. Huma.ai. And I'm on LinkedIn. And just Google, just search my name. It will come up. There you go. Thank you very much for coming to the show. We really appreciate it. Thank you for having me. This has been fun. There you go. Thanks to my audience for tuning in. Go to goodreads.com, FortressCrispFoss, LinkedIn.com, FortressCrispFoss, and YouTube.com, FortressCrispFoss.

Be good to each other. Stay safe. And we'll see you guys next time. And that should have...