Transforming Healthcare with AI: How Dandelion Health is Leading the Way with Elliott Green, co-founder and CEO of Dandelion Health

[00:00:02] [SPEAKER_00]: Hey everyone, welcome back to the Outcomes Rocket founder stories. So excited that you

[00:00:08] [SPEAKER_00]: tuned into our podcast again because today I've got the privilege of hosting Elliott Green.

[00:00:14] [SPEAKER_00]: He is the co-founder and CEO of Dandelion Health. They're a real world data and clinical

[00:00:22] [SPEAKER_00]: AI platform powering the next generation precision medicine and personalized care.

[00:00:28] [SPEAKER_00]: His career has spanned finance and health technology, culminating in executive roles

[00:00:34] [SPEAKER_00]: within this tech health area. They're really focused on payers, providers, life sciences,

[00:00:40] [SPEAKER_00]: and healthcare data. I'm excited to have them here on the podcast. Elliott, welcome.

[00:00:46] [SPEAKER_00]: So thanks so much for the invitation. Pleasure to be here.

[00:00:49] [SPEAKER_00]: Hey listen, it's our pleasure to host you. And today we're going to have a chance to

[00:00:53] [SPEAKER_00]: really unpack the work that you and the Dandelion Health team are up to. But before

[00:00:57] [SPEAKER_00]: we do, tell us about you. What is it that drives your mission in healthcare and also

[00:01:02] [SPEAKER_01]: entrepreneurship? Yeah, a nice big topic to start with.

[00:01:08] [SPEAKER_01]: I'll start with the healthcare bit first. I was one of these people that was kind of

[00:01:13] [SPEAKER_01]: lucky enough to kind of fall into healthcare. You can probably tell being English, we have

[00:01:18] [SPEAKER_01]: a very different health system. And so the complexity of the US healthcare system

[00:01:23] [SPEAKER_01]: was astonishing to me once I'd kind of got into it. And I think once you do,

[00:01:28] [SPEAKER_01]: and you kind of get bitten by the bug as so many people say, it's that two-sided coin of,

[00:01:33] [SPEAKER_01]: my God, this feels like a sysophane task. But oh, if we could just manage this bit,

[00:01:39] [SPEAKER_01]: the impact we could have on so many people is unlike anything you could ever do.

[00:01:44] [SPEAKER_01]: And so that allure of trying to be this impactful in such a positive fashion in

[00:01:49] [SPEAKER_01]: society and health is really what keeps you going on a personal level. I just kept on having

[00:01:54] [SPEAKER_01]: experiences that taught me that the more you could kind of control and move towards as an

[00:02:00] [SPEAKER_01]: entrepreneur, the bigger the impact could be. And so I was more of a, I would say cumulative

[00:02:06] [SPEAKER_01]: journey. Some people kind of wake up in the morning, you have their bowl of shreddies and

[00:02:10] [SPEAKER_01]: they're like, I want to start a company. And mine was a bit more, I got lots of experiences

[00:02:15] [SPEAKER_01]: through Oscar, through Charles Bark and through Clarify and realized, wait a second,

[00:02:19] [SPEAKER_01]: if I bring all of this together, I could do something really impactful with a great team.

[00:02:24] [SPEAKER_01]: And that's also the key thing, right? If you don't have a good team, it doesn't

[00:02:27] [SPEAKER_01]: matter how much of an entrepreneur you are, you will achieve nothing.

[00:02:30] [SPEAKER_00]: Amen to that, Elliot. So true. And so this is super fascinating. So it's an area of high

[00:02:37] [SPEAKER_00]: impact for you. You decided to go for it. Tell us about dandelion. What is it that

[00:02:41] [SPEAKER_01]: you guys are up to? So as you said in the intro, real world data platform focused on

[00:02:46] [SPEAKER_01]: clinical AI and kind of precision analytics. Great buzzwords. What does it really mean?

[00:02:51] [SPEAKER_01]: What makes us very different is that we looked at the world of kind of AI and my co-founder's

[00:02:56] [SPEAKER_01]: a kind of world famous in that space of AI and healthcare. My other ones are leading

[00:03:00] [SPEAKER_01]: health system exec and between all of us, we had different perspectives. And what we

[00:03:04] [SPEAKER_01]: universally understood was that AI was in an interesting dilemma of having a data problem

[00:03:09] [SPEAKER_01]: more than an implementation problem. Often in healthcare, it's, oh, I don't want to

[00:03:13] [SPEAKER_01]: alter the physician workflow or how do I get an insurance company to pay for this? And all of that

[00:03:17] [SPEAKER_01]: stuff is going to happen. But ultimately, the beginning of AI, it's a huge data issue of how

[00:03:23] [SPEAKER_01]: do I get hold of the best quality data to build the right products? And that was where you take

[00:03:29] [SPEAKER_01]: open AI and LLMs, where you had the internet. It was very easy. You're just getting harder.

[00:03:34] [SPEAKER_01]: Obviously, it was people steep on putting gates behind things. But ultimately,

[00:03:37] [SPEAKER_01]: you could train on this huge data set. In healthcare, that's not the case. It's hard to get past that

[00:03:42] [SPEAKER_01]: walled garden into the really interesting stuff, the MRIs, the CTs, the lab tests, the outcomes,

[00:03:48] [SPEAKER_01]: like all of that together. That's what you want. And so that's where we got to was understanding

[00:03:54] [SPEAKER_01]: that we then realized, well, health systems hold this information. Let's build a company that

[00:03:59] [SPEAKER_01]: partners with those health systems, the right kind of health systems that have that data,

[00:04:03] [SPEAKER_01]: pull out this high quality data, curate it, harmonize it and make it available for really,

[00:04:09] [SPEAKER_01]: really smart people to go off and build products that would ultimately change the face of healthcare.

[00:04:15] [SPEAKER_01]: And as we got into that, we've been doing that. So we extract petabytes of data from health systems,

[00:04:20] [SPEAKER_01]: de-identify it, move it into the cloud, harmonize it. And then we have life sciences

[00:04:23] [SPEAKER_01]: companies, medical device companies, AI developers use it. And we try and further science

[00:04:28] [SPEAKER_01]: wherever we can in the hopeful pursuit of increasing quality of outcomes.

[00:04:33] [SPEAKER_00]: That's fascinating, Elliot. Yeah. And so what do you find today is the biggest use case and opportunity?

[00:04:42] [SPEAKER_01]: So about two months ago, we launched a GLP-1 library.

[00:04:45] [SPEAKER_01]: Oh, that's interesting.

[00:04:46] [SPEAKER_01]: And yeah, there's...

[00:04:47] [SPEAKER_01]: That's a hot item.

[00:04:50] [SPEAKER_01]: And as it was funny, the reason we did this was because kept on reading these articles

[00:04:54] [SPEAKER_01]: about people saying, well, GLP-1s, they could have all this incredible impact. And

[00:04:58] [SPEAKER_01]: you realize historically, that would normally take potentially decades and hundreds of millions,

[00:05:03] [SPEAKER_01]: if not billions of dollars to find out. And actually, that impact is happening right here and now.

[00:05:09] [SPEAKER_01]: People are on GLP-1s and they're seeing their cardiovascular risk or decline. They're seeing

[00:05:13] [SPEAKER_01]: their liver fat decrease. They're seeing then your... Maybe changes in their neurology.

[00:05:18] [SPEAKER_01]: It's just we can't access the data to assess it. The only way you can access that data

[00:05:22] [SPEAKER_01]: is normally by running a clinical trial and very specifically getting people to give you

[00:05:26] [SPEAKER_01]: that data. So you can analyze it, that takes years. But if you can find that data now happening

[00:05:30] [SPEAKER_01]: to patients within these health systems and across America and you make it available to smart people,

[00:05:37] [SPEAKER_01]: you can actually find out right now. So to give you an example, in a very short while,

[00:05:41] [SPEAKER_01]: we'll be coming out with a study that actually looked at cardiovascular risk in GLP-1s

[00:05:46] [SPEAKER_01]: and used ECGs. And we were able to discover this population that was completely novel to

[00:05:53] [SPEAKER_01]: anything that had been seen in a clinical trial because using artificial intelligence,

[00:05:58] [SPEAKER_01]: you were able to extract this data from an ECG and understand the impact of a GLP-1

[00:06:03] [SPEAKER_01]: in a way that no one had ever done before. And that's where this combination,

[00:06:07] [SPEAKER_01]: these two incredible fields of get the right data, this huge sway of real world data with

[00:06:12] [SPEAKER_01]: like raw biological information and AI, that's the kind of paradigm shifting combination.

[00:06:17] [SPEAKER_00]: Yeah, no, I definitely love that. So with this particular subset of patients in cardiology,

[00:06:24] [SPEAKER_00]: what was the insight? Like what was the big aha?

[00:06:26] [SPEAKER_01]: Oh, I don't know if I can give that to you, Saul. It's two weeks early. But basically,

[00:06:30] [SPEAKER_01]: oh, is it?

[00:06:30] [SPEAKER_01]: Yeah, we're coming out with a really sick. But ultimately, it's that the GLP-1s had a far

[00:06:36] [SPEAKER_01]: more positive effect and impact on a different strata of the population than people had

[00:06:41] [SPEAKER_01]: even studied. Clinical trials will focus on a certain strata of population, right? Because

[00:06:47] [SPEAKER_01]: that's what you need to prove efficacy. Whereas AI is able to analyze people outside of those

[00:06:52] [SPEAKER_01]: trials because you can just take their raw data. And so what's this space? I will send you a copy

[00:06:57] [SPEAKER_01]: of the white paper as soon as we're down. But we think it's incredibly impactful, because

[00:07:01] [SPEAKER_01]: that's the point, right? What you want to do is not just get the white elephant of all of this

[00:07:05] [SPEAKER_01]: data. It's you want to turn into something actionable and somebody is going to benefit

[00:07:09] [SPEAKER_01]: society. And that's where I think that combination of AI and data will be so amazing.

[00:07:14] [SPEAKER_00]: That's awesome. Yeah, no, no. I love it. And you know, I love that you went for GLP-1s just because

[00:07:20] [SPEAKER_00]: it's a hot topic. It's costing a lot of money. There's a lot of questions. What are we doing here?

[00:07:27] [SPEAKER_00]: And yeah, like rather than be restricted to a research cohort, the data is out there. Let's

[00:07:33] [SPEAKER_00]: do something with it. So I think it's really cool, man. And I was at the American College of

[00:07:38] [SPEAKER_00]: Cardiology this year, big topic GLP-1s, but also the use of AI. And this is just a great example of

[00:07:45] [SPEAKER_00]: things you could do with it. So I appreciate the real use case, like tangible use case. Talk to us

[00:07:51] [SPEAKER_00]: about what makes you guys different, right? There's a lot of players out there in the space

[00:07:56] [SPEAKER_00]: that are trying to make an impact with data and AI. Sir, I mean, fundamentally, one of the

[00:08:01] [SPEAKER_01]: big things that makes us different is we spent years building pipes to do this properly.

[00:08:06] [SPEAKER_01]: Yeah, I could come out and say, I've got 300 million lives. It's this incredible real-world

[00:08:10] [SPEAKER_01]: data. It will be skimming the surface. What you really want to do this effectively or sending

[00:08:14] [SPEAKER_01]: the way that we're trying to is you want all of those raw images, raw ECGs, raw data, that takes

[00:08:19] [SPEAKER_01]: years to work that out with health systems to be able to get into the data and extract it,

[00:08:25] [SPEAKER_01]: de-identify it, do it ethically and responsibly. And so there's no substitute for that. It

[00:08:29] [SPEAKER_01]: doesn't matter how much AI revolution. Like that's just nuts and bolts, right? Pure plumbing.

[00:08:33] [SPEAKER_01]: And so that's the first thing, like we just spent a long time doing it. Secondly, we work with

[00:08:38] [SPEAKER_01]: these health systems because the curation of the data, the context of the data is so key.

[00:08:43] [SPEAKER_01]: I could go and get a lot of data, but I wouldn't necessarily understand it. It could be a perfect

[00:08:48] [SPEAKER_01]: example. There are like 17 discharge dates in one of our health systems, EMRs. And the reason

[00:08:53] [SPEAKER_01]: there are 17 discharges is that it discharges all at different points in the system. Is it

[00:08:57] [SPEAKER_01]: the ICU to the main floor? Is it the ED to the ICU? Did you actually go home? Did what we do?

[00:09:02] [SPEAKER_01]: And so unless you've got context over something that simple, you don't necessarily know exactly

[00:09:09] [SPEAKER_01]: what the outcome was. And so that's the other thing. We plan with our health systems to get

[00:09:12] [SPEAKER_01]: that information so that we can have the highest fidelity data and then speed because we've got

[00:09:18] [SPEAKER_01]: that piping bill. We've done the curation properly. Now what we can do is we can move

[00:09:22] [SPEAKER_01]: really fast to getting the insights. And so that combination is awesome. And then we also

[00:09:26] [SPEAKER_01]: built this validation for AI capability. So what we actually do is we take this really diverse

[00:09:32] [SPEAKER_01]: population. So we built our health systems with three health systems, Sharp in San Diego,

[00:09:36] [SPEAKER_01]: Sanford in the Dakotas, largest rural health system in the US, and Texas Health in Dallas.

[00:09:41] [SPEAKER_01]: And we've got a fourth coming in the northeast. And that takes us to about 15 million patients.

[00:09:45] [SPEAKER_01]: And it's completely a cross section of race and ethnicity in the US, as you would imagine.

[00:09:49] [SPEAKER_01]: And so what that allows us to do is it allows us to take algorithms and validate them

[00:09:53] [SPEAKER_01]: on this data and assess them to say, are you actually performing as you should number one?

[00:09:59] [SPEAKER_01]: And two, is there any bias? Are you accidentally biased against Hispanic ladies between 40 and 55

[00:10:05] [SPEAKER_01]: because guess what? They just weren't in your training set. And so that combination of all

[00:10:10] [SPEAKER_01]: of those things means that we can take this high quality data, high quality algorithms

[00:10:15] [SPEAKER_01]: and create a high quality product that can ultimately be used to benefit.

[00:10:20] [SPEAKER_00]: Interesting. Now that's really great. So let's talk about stakeholders. So you've got health systems

[00:10:26] [SPEAKER_00]: that have the data. And then you mentioned the industry stakeholders like Pharma, MedDevice.

[00:10:33] [SPEAKER_00]: So what is the provider looking for? Right? When they're saying, here's my data,

[00:10:40] [SPEAKER_00]: but like what's in it for me as a provider?

[00:10:42] [SPEAKER_01]: That's a great question. The key thing for us and I think you'll appreciate this,

[00:10:46] [SPEAKER_01]: having partnerships are all about how do I align incentives? Not in a full sway, but in a real long

[00:10:52] [SPEAKER_01]: term, like if it's not win, win, win kind of thing for everyone, it just won't work.

[00:10:56] [SPEAKER_01]: Right. And so how we designed Dandyline was the systems are our stakeholders,

[00:11:00] [SPEAKER_01]: but so are the patients. And so ultimately, if you think about it,

[00:11:04] [SPEAKER_01]: if we didn't do this with our partners, how would this work? Well, probably what would

[00:11:08] [SPEAKER_01]: happen is Google, Amazon, a bunch of other companies would get all of this data. They'd

[00:11:11] [SPEAKER_01]: create a bunch of solutions. Maybe they'd be good. Maybe they wouldn't. Who knows?

[00:11:14] [SPEAKER_01]: And then they kind of be foisted onto the healthcare systems. They wouldn't really be

[00:11:18] [SPEAKER_01]: involved in it, be involved in the product development. They just kind of have to use it.

[00:11:22] [SPEAKER_01]: And we've seen that play out and it's never as pretty as one would like, right?

[00:11:26] [SPEAKER_01]: Right. Whereas this way, the health systems can be involved in the actual development. Like

[00:11:31] [SPEAKER_01]: they can consult with some of the companies, they can be involved, like we're going to run

[00:11:34] [SPEAKER_01]: clinical trials at some of them. And then also they can add their expertise.

[00:11:38] [SPEAKER_01]: And so their patients will get those solutions earlier and other health systems around the

[00:11:44] [SPEAKER_01]: country will know that providers were involved in some of the things that we're doing a lot

[00:11:48] [SPEAKER_01]: earlier that would ultimately lead probably to a better outcome. So that's for the provider,

[00:11:52] [SPEAKER_00]: I think, where the biggest benefit is. And so is it about, I mean, are we squaring this into

[00:11:58] [SPEAKER_00]: the life sciences sector primarily and kind of quicker drugs to market,

[00:12:03] [SPEAKER_00]: right treatments to the patients when they need it? That type of value?

[00:12:06] [SPEAKER_01]: I think in the short term, where we've found the industry that's most willing to take AI on board

[00:12:12] [SPEAKER_01]: is going to be the life sciences. You know, like life sciences and med device are often a

[00:12:18] [SPEAKER_01]: little further up the curve in terms of adoption of these things, right? And so

[00:12:21] [SPEAKER_01]: one of the things that we'll find is payers and providers will come later. But yeah, life

[00:12:25] [SPEAKER_00]: science is a med device initially. No, I love it. Super interesting and valuable.

[00:12:30] [SPEAKER_00]: And there's, you know, like on the devices side, I spent 17 years on the devices side,

[00:12:35] [SPEAKER_00]: there's so much data that those devices produce. So like let's throw on a med device hat.

[00:12:44] [SPEAKER_00]: What do you want to tell them? Like, you know, you guys are generating all this data with

[00:12:48] [SPEAKER_00]: your devices. What do you want to tell them about that? So this is really, really interesting.

[00:12:53] [SPEAKER_01]: A lot of people would say, okay, look at Philips, Metronik and all these guys,

[00:12:56] [SPEAKER_01]: why would they ever need you? They've got so much data. Surely this has already been done,

[00:13:01] [SPEAKER_01]: but you'll know this from having worked there. You have data on exactly what happens when somebody's

[00:13:06] [SPEAKER_01]: using a device, but no context outside of that. Don't know how they got to wearing it. You don't

[00:13:11] [SPEAKER_01]: know what happens to them afterwards. And so how can you improve a device if you don't really

[00:13:16] [SPEAKER_01]: understand the impact that it's had or what it's measured or what the outcome was. And so

[00:13:20] [SPEAKER_01]: for med device, one of the key things is we have this longitudinal data.

[00:13:24] [SPEAKER_01]: Take an ECG, right? It's a perfect example. So we'll walk into the ED with the chest pain,

[00:13:30] [SPEAKER_01]: the physician will look at the ECG. They probably can only tell if there's an arrhythmia.

[00:13:34] [SPEAKER_01]: And then they make a decision whether or not that person with chest pain should go and have

[00:13:38] [SPEAKER_01]: an echocardiogram or should go home because they think it was indigestion, but they don't know.

[00:13:41] [SPEAKER_01]: They're kind of phenotyping and guessing. And Philips and Metronik can't do anything

[00:13:45] [SPEAKER_01]: about that, right? You can't change that. You're not going to be able to get AI to

[00:13:50] [SPEAKER_01]: diagnose that because you can only use a human. And the human can only see if it's an arrhythmia.

[00:13:55] [SPEAKER_01]: If I got longitudinal data and I see outcomes, now I can see, okay, these are the patients

[00:13:59] [SPEAKER_01]: who had this waveform. And then these are the ones that got discharged and were fine.

[00:14:03] [SPEAKER_01]: And these are the ones that sadly had a cardiac arrest. And these are the ones that had this

[00:14:06] [SPEAKER_01]: and had that. And you can train an algorithm to recognize those patterns. And now you

[00:14:10] [SPEAKER_01]: can adjust the software on your device to say instead of when the ED physician looks at the

[00:14:16] [SPEAKER_01]: ECG and has to guess, now they have a run-down based on an algorithm you have built to say,

[00:14:21] [SPEAKER_01]: there's an 84% chance you should send this person through to an echocardiogram now based on X, Y,

[00:14:26] [SPEAKER_01]: and Z. But only with longitudinal data and outcomes can you build that world.

[00:14:31] [SPEAKER_00]: Yeah. It's an interesting thing to think about, right? The ecosystem that our devices,

[00:14:36] [SPEAKER_00]: the devices that the manufacturers, you're listening to this, right? Your devices,

[00:14:40] [SPEAKER_00]: they're in the ecosystem. So where do they sit with regard to that longitudinal journey

[00:14:45] [SPEAKER_00]: of the patient? It's critical that we think about how things are feeding in and potentially,

[00:14:51] [SPEAKER_00]: I don't know, I mean, does this sort of potentially inform pipeline for their development?

[00:14:57] [SPEAKER_01]: Dashboards. I mean, let's be honest, everyone is trying to work out how to utilize AI.

[00:15:02] [SPEAKER_01]: I mean, even the LLMs, they're trying to work out. So healthcare is no different.

[00:15:07] [SPEAKER_01]: Healthcare just has this added complexity of, not only do I have to work out what it does,

[00:15:11] [SPEAKER_01]: but who pays for it? Which is always the classic. But there are so many... This is why AI is going

[00:15:16] [SPEAKER_01]: to be such an interesting shift because it's so clear that you can not only create value,

[00:15:22] [SPEAKER_01]: but you can actually understand who that creation of value, that value of creation,

[00:15:26] [SPEAKER_01]: who it goes to. So to give you an example, we're looking at an algorithm now where we're

[00:15:30] [SPEAKER_01]: assessing chest CTs and emphysema. And the algorithm predicts emphysema. And so what we

[00:15:35] [SPEAKER_01]: did was we compared it to real life and we said, okay, well, this is what real life looked like. Now

[00:15:40] [SPEAKER_01]: we use the algorithm. This is the delta between the two. When did the algorithm catch it? How

[00:15:45] [SPEAKER_01]: much earlier? How much more accurate was it? But also then you can start working out,

[00:15:49] [SPEAKER_01]: is it better for fee-for-service? Is it better for value-based care? Is this a payout product?

[00:15:54] [SPEAKER_01]: Is this a provider product? And that starts to help medical device AI developers understand

[00:16:00] [SPEAKER_01]: what you're building and who you're building it for. And only data can allow you to have that

[00:16:06] [SPEAKER_01]: insight. And we haven't had access to that kind of data before. So that's really where the big shift

[00:16:11] [SPEAKER_00]: is. Love that. Super interesting conversation, Elliot. Really enjoying this and the value is

[00:16:16] [SPEAKER_00]: clear. Let's shift to the building of a company. We learn more from our setbacks than our

[00:16:23] [SPEAKER_00]: successes sometimes. What setback would you point to that's taught you a ton about the business and

[00:16:30] [SPEAKER_01]: it has helped you make it that much better? Good question. People is always one. I'll give

[00:16:35] [SPEAKER_01]: you that in a second. I think the other big one is work out when you need plan Bs. You don't

[00:16:42] [SPEAKER_01]: always, but you need to understand where your key risks are. And so the setback we had was we

[00:16:47] [SPEAKER_01]: got to the one-yard line like a two-year process with the hospital system. Thought we

[00:16:51] [SPEAKER_01]: were about to get the deal done for the data. And then suddenly they had a change of like process.

[00:16:56] [SPEAKER_01]: And I got learned two things. One, I didn't have a backup at that particular moment. It was hard

[00:17:00] [SPEAKER_01]: to do it. But two, and so I started to think differently. And also how do I defeat, how do

[00:17:04] [SPEAKER_01]: I like take that risk and bring it up front? How do I make sure I'm seeing around more corners

[00:17:09] [SPEAKER_01]: so that I don't get surprised on the one-yard line? I like know about it way, way before

[00:17:13] [SPEAKER_01]: that, right? So that was one. Two, I actually stuck in there and now they're part of the

[00:17:17] [SPEAKER_01]: consortium. So, you know, worked out. And so that was a big one. And then I'd say on people,

[00:17:23] [SPEAKER_01]: you know, when I started this, you always hear the classic of the people you hire,

[00:17:28] [SPEAKER_01]: the quality of the people is the single most important thing. And I think you

[00:17:31] [SPEAKER_01]: get that, but you don't really, really get it until you have to go through that scaling up the

[00:17:36] [SPEAKER_01]: first time. And it is about attitude. It's about flexibility. It's also about the fact

[00:17:43] [SPEAKER_01]: that you need people that you don't have to direct at the beginning on every level thing

[00:17:47] [SPEAKER_01]: and that you can give them a task. And, you know, that classic thing of exceed expectations.

[00:17:53] [SPEAKER_01]: You know, what's been so wonderful about Dandelion and the amazing team that I've been

[00:17:56] [SPEAKER_01]: fortunate enough to create is each time I give somebody something or they look at it,

[00:18:00] [SPEAKER_01]: it's always an exceeding expectation that comes back to me better than I would have

[00:18:04] [SPEAKER_01]: thought it was going to. That's awesome. And that's an amazing thing.

[00:18:07] [SPEAKER_00]: Yeah, for sure. And man, loud and clear on the unforeseen events, they're always going

[00:18:12] [SPEAKER_00]: to happen, whether it's a deal and congrats on getting it anyway and making that happen.

[00:18:18] [SPEAKER_00]: But the timing, right? Like so like whether it's a deal, whether it's funding,

[00:18:22] [SPEAKER_00]: whether it's a partnership, like just thinking about those corners, like looking around those

[00:18:27] [SPEAKER_00]: corners and having that plan B is such a great tip for all of us to keep in mind. So Elliott

[00:18:32] [SPEAKER_00]: really appreciate that insight for us. It's all about making an impact

[00:18:37] [SPEAKER_00]: with each podcast and with each founder that we have on the podcast. Leave us with what you want

[00:18:43] [SPEAKER_00]: all of our listeners and viewers to be thinking, a call to action there, and then the best place

[00:18:49] [SPEAKER_00]: that they could reach out to you to learn more about what you and the Dandelion team are up to.

[00:18:54] [SPEAKER_01]: It's hard not to continue the theme of AI. I mean, I could talk about employer-led health

[00:18:58] [SPEAKER_01]: insurance all day, but that's not where we're going. I think people have a vague sense of

[00:19:03] [SPEAKER_01]: impactful AI would be. I don't think that there are enough tangible examples of really understanding

[00:19:08] [SPEAKER_01]: it. And so I think a couple of ones that would be really good to leave you with as it were.

[00:19:13] [SPEAKER_01]: I think of AI almost like this hyperpowered physician, right? In an ideal world, what would

[00:19:18] [SPEAKER_01]: you do? Your loved ones in the hospital bed, you're standing there and the physician comes in

[00:19:22] [SPEAKER_01]: the attending having like hurriedly read maybe a page of probably 400 pages of notes.

[00:19:28] [SPEAKER_01]: And so the classic thing would be the LLM can like summarize it and they can read them quickly

[00:19:33] [SPEAKER_01]: and that will help. But it's also all the other information. It's the MRI and the CT and the ECG

[00:19:39] [SPEAKER_01]: and the other diagnosis. What you want is this hyperpower to take all of that information and

[00:19:44] [SPEAKER_01]: distill it into a really high-powered diagnosis, right? That would be amazing. And similarly,

[00:19:50] [SPEAKER_01]: and that's what AI can do, right? It does what humans can't, it augments that capability

[00:19:54] [SPEAKER_01]: and then humans stay in the loop to assess whether or not that was accurate and if you've

[00:19:58] [SPEAKER_01]: learned something. And then the life sciences side, I think this really incredible bit of

[00:20:04] [SPEAKER_01]: drug development right now takes far too much money and takes far too much time.

[00:20:09] [SPEAKER_01]: And over COVID, the FDA got a really good experience of using real-world evidence

[00:20:13] [SPEAKER_01]: to kind of look at the vaccines and understand how you could do some of that.

[00:20:17] [SPEAKER_01]: And I think the more that this high quality multimodal data comes out,

[00:20:21] [SPEAKER_01]: the less crazy it is to imagine broader, more impactful, faster, cheaper trials

[00:20:28] [SPEAKER_01]: really helping to propel drug development in a way that you know, it hadn't done previously.

[00:20:33] [SPEAKER_01]: Even down to like label expansion, you know, there are lots of treatments out there where

[00:20:37] [SPEAKER_01]: you often get your physician will say, oh, I've heard that this could be really good for

[00:20:41] [SPEAKER_01]: this off-label. But wouldn't it be amazing if you had the data that could actually

[00:20:44] [SPEAKER_01]: securely tell you that information and could say actually it's really good for

[00:20:49] [SPEAKER_01]: this particular population. And that matches you. And I've got the evidence here to show it.

[00:20:54] [SPEAKER_01]: That's where we should be going towards, right? That personalized care.

[00:20:57] [SPEAKER_01]: And so I think that's where we're starting and for the first time ever. And I think

[00:21:02] [SPEAKER_01]: that's where we'll get to.

[00:21:03] [SPEAKER_00]: That's awesome Elliot. Now I love the very clear picture you've painted

[00:21:07] [SPEAKER_00]: for the future of AI and healthcare as an augmentation tool, but also as a tool

[00:21:13] [SPEAKER_00]: that helps really reduce the amount of time and investment that it takes to get drugs

[00:21:17] [SPEAKER_00]: to market to the right people at the right time. Tell us where people can reach out to you.

[00:21:21] [SPEAKER_00]: Like let us know what's the best place to get in touch.

[00:21:24] [SPEAKER_00]: So I probably a bit old at this point, it's still linked in for me is probably the best way.

[00:21:28] [SPEAKER_01]: Yeah, it's great. It's good. I'll be a health. I'm on a panel.

[00:21:32] [SPEAKER_01]: I'll see you there then.

[00:21:33] [SPEAKER_01]: I'll see you there.

[00:21:33] [SPEAKER_01]: We'd love to talk to anyone that's going to be around the ones that's okay

[00:21:36] [SPEAKER_01]: and healthcare and then come to the website and anyone who's interested in this space.

[00:21:41] [SPEAKER_01]: We're always looking to have good conversations with people and see what happens.

[00:21:46] [SPEAKER_00]: Outstanding. This has been a lot of fun Elliot. Thank you so much for joining us folks.

[00:21:51] [SPEAKER_00]: Elliot Green, co-founder and CEO of Dandelion Health, bringing that real world data and clinical AI

[00:21:59] [SPEAKER_00]: all within this platform that gives us that precision medicine and personalized care

[00:22:04] [SPEAKER_00]: that we're all looking for. Reach out to them. We're going to leave their

[00:22:07] [SPEAKER_00]: contact information in the show notes. So make sure you check those out to get

[00:22:11] [SPEAKER_00]: in touch with Elliot. That's how you make stuff happen. Elliot, I want to thank you

[00:22:15] [SPEAKER_00]: for joining us. This has been a lot of fun.

[00:22:17] [SPEAKER_00]: Now the pleasure was on my hands so thanks so much again.