Please enjoy this transcript of my conversation with Emi Gal, Founder & CEO of Ezra. Transcripts for other episodes can be found here.
Transcript – #146 Ezra: Bringing Fast and Affordable Early Cancer Detection to Everyone | Emi Gal, Founder & CEO
Daniel Scrivner (00:00:00):
Hello, and welcome to another episode of Outlier Founders by Outlier Academy, where we decode what iconic founders and entrepreneurs have mastered from how they've built their companies to the frameworks and strategies they use, and the lessons they've learned along the way.
I'm Daniel Scrivner, and on the show today I'm joined by Emi Gal. Emi is the founder and CEO of Ezra, where he is focused on bringing affordable and fast cancer screening to everyone. Since founding Ezra in 2018, Ezra has managed to bring down the cost of MRI cancer screening by 80%, and the time required to get your body scanned by 66%, and they're not done. They think they can get it to nearly 90%, all by harnessing the power of AI and machine learning.
Listen as we decode everything from the incredible technology behind Ezra's approach to cancer screening, to why magnetic resonance imaging or MRI is the right technology to screen for cancer, how Ezra has built a capital A business in deep tech healthcare, and why the cure for cancer already exists. It's just not evenly distributed yet. You can find a searchable transcript of this episode, as well as our episode guide, with ways to dive deeper into the topics we cover at outlieracademy.com/146. That's outlieracademy.com/146. Please enjoy my conversation with Emi Gal.
Emi, thank you so much for joining me on Outlier Founders. I'm thrilled to have you on to talk about Ezra.
Emi Gal (00:01:19):
Thanks for having me, Dan. It's great to be here.
Daniel Scrivner (00:01:21):
So we're going to cover a lot of ground today, and shortly we will break down what Ezra is, but I want to start with your background, because you have, I think, a pretty particularly interesting background. And obviously, Ezra is the second company that you've started. So give us just a quick sketch of your background, kind of up to founding Ezra.
Emi Gal (00:01:40):
Absolutely. So I'm originally from Romania. I studied computer science and applied mathematics in Bucharest. And while I was in college, I started a software company, I basically didn't have any money, wanted to make some money, and started a software company with a bunch of friends in college who were also programmers. And initially, we just started building software for others in the US and Europe. That turned out to be a really good business at the time. And then we used the cash to launch our own product, and we had this idea to launch a video ad server, which back in 2008-2009 was a great idea, because video was very new on the internet.
That company led me to London. I spent seven years in London, and then it was acquired by a large ad tech group in New York City. And the reason I'm giving you that context is because of my experience with [inaudible 00:02:33], I made some money, I started volunteering for a non-profit in Romania called Hospices of Hope, who build and operate hospices that cared for cancer patients.
And in being involved with them, I realized that the main reason why people end up in a hospice due to cancer is because they found cancer late. And the main reason we still find cancer late for people is because there's no way to screen for cancer everywhere in the body, that's fast, accurate, and affordable. And so, decided to focus on that problem, and had some cancer experience in my own family. My mother sadly passed away from cancer, so kind of very, very motivated by this mission, and went on to do a couple of years of research and then start Ezra.
Daniel Scrivner (00:03:22):
I want to talk about those couple of years of research, because you did something relatively unique there, where you obviously identified a problem, or the way I like to think about it, you identified a very large and interesting problem space, you had a very clear idea of what you wanted to do, or where you needed to move the needle in order to change some of the outcomes, but then you went through this process of an exhaustive list of the different ways you could solve the problem, and trying to eliminate them. Talk about that, and why you landed on the approach that you landed on with Ezra.
Emi Gal (00:03:54):
Yeah. So Ezra in the current iteration was the 12th idea or iteration of how you could go about screening for cancer. And after setting on this as the problem, I basically went through all of the different ways in which you could attack the problem. And I had some crazy ideas like DNA-based nanobots that you put in the bloodstream, that would kind of beam data to a Bluetooth device and whatnot.
Daniel Scrivner (00:04:24):
I mean, that sounds fascinating for like 2040.
Emi Gal (00:04:26):
And I actually met with a couple of scientists who were working on DNA-based nanobots, and concluded very quickly that the technology is way too early to be able to use it today. And really my goal, or one of my criteria for choosing a solution was it has to be something that can be used today, where with technology, you could make it better and more accessible. And so, I cycled through about a dozen ideas. And then actually, on my honeymoon, this is November of 2016, December of 2016, I was reading a paper that was comparing MRI scans with other types of imaging, such as CT scans, ultrasound, and so on, in terms of the sensitivity and specificity for detecting cancer early. And it was concluding overwhelmingly that MRI has higher sensitivity, high specificity, and on top of that, it doesn't expose you to ionizing radiation.
So settled on MRI as one of the solutions that you could apply to detect cancer early. However, MRI has a big issue, which is it uses magnetic resonance of the protons in the water of your body to create internal images of the body. And because it uses magnetic resonance, it's very noisy. And so, to do an MRI scan, you generally need to do the same scan multiple times because signal will always be the same, noise will always be random, and by doing it multiple times, you can average out the noise, the randomness of the noise. And so, with my computer science hat on, I was like, "Well, we can probably create machine learning models that identify what noise looks like, and remove that noise without having to do the scan multiple times."
So decided that MRI can be applied to this, built a prototype initially focused on prostate cancer screening, using MRI, launched that, it was very successful. And then we launched our full-body MRI in 2019, and we've since scanned a few thousand people, and found potential cancer in 13% of our members. Almost every week we find something for members, right now.
Daniel Scrivner (00:06:38):
Wow. I mean that's amazing outcome, just even the size of that already with the number of patients that you're treating, the number of customers that you're treating. So just to try to parrot that back, because it is a lot to wrap your head around. As I had initial conversation with you when I was preparing for this, there's a lot here. So just to kind of pair it back, Ezra is, you're focused on cancer screening, cancer detection early. You're trying to bring down the cost, and obviously increase the accessibility. Kind of cost is a major barrier to accessibility. And then you're taking an existing technology that's MRI, and then you're kind of adding onto that obviously a software layer that's AI, using a lot of machine learning. And you're using that to effectively, I'm guessing, up-res the imagery in some regards, or strip out some of the noise?
Emi Gal (00:07:21):
Yeah. So we actually apply AI to all of the spectrum of screening. So there are three big cost centers to getting screened with the full body MRI. There is the scan itself, there is the radiology interpretation, and there is the report generation for the person to actually understand what the interpretation means. And actually, Ezra use AI across that entire spectrum, make the MRI scan faster, make the interpretation faster, and make the report generation faster. In our case, the report generation is actually automated. We take radiology reports, and we, automatically using ai, turn them into Ezra reports that then get reviewed by a medical provider, and sent to members. So our thinking is that machine learning can be used to decrease all of the cost centers, or for all the cost centers in order to make our scan more affordable, and then pass all of those cost savings to consumers.
Daniel Scrivner (00:08:22):
Yeah. I want to talk about one of the things I find very helpful is just comparing and contrasting, and we're going to go into, later in this interview, aspects of your model that I think are really interesting, that are effectively allowing you to take this MRI technology, make it available in a way that it's never been available, make it affordable in a way that it's never been affordable. And so, I think it'd be helpful to just start with a comparison. And so, obviously you talked about that, today, cancer screening almost sounds reactive, it's something that you don't do proactively. And so typically, you're feeling ill, or maybe just randomly for something else, they're doing an MRI, and they end up finding cancer. Talk about, I guess, I don't know, or paint a picture of today how it would work to try to screen for cancer, if someone's not using Ezra or what that process usually looks like?
Emi Gal (00:09:08):
Yeah. So every year, about 20 million people will be diagnosed with cancer globally. Half of these individuals will be diagnosed late. This means that they have a symptom, they went to the doctor, the doctor said, "Oh, let's do a scan." And they would do this scan, and they had metastatic colon cancer with expansion to the liver. And that's 10 million people every year. Now, the reason this happens is because about 50% of cancers don't have any screening guidelines.
If you're fortunate enough to get, and I say fortunate, quote unquote, but if you're fortunate enough to have breast cancer, you will likely be able to find it early, because you can get a mammogram every year, if you're a woman, which is very good for finding breast cancer early. But if you are unlucky enough to get pancreatic cancer, you're not going to know until you're symptomatic, and you'll only be symptomatic most of the time when it's expanded beyond pancreas. And that means that your five-year survival rate will be in the single digit percentages.
Now, what we're saying at Ezra is actually there are ways to find these cancers early. And actually, if you pull together the prevalence for all of the cancers we screen for at Ezra, we have similar prevalence for people over 40 as breast cancer or lung cancer. Breast cancer is, I don't want to say 20% prevalence in women over 40, if you pull together the 13 organs we screen for at Ezra, we have a 20% or so prevalence. And so, what we're trying to do is to find cancer early, because five-year survival rate for early stage cancer is about 80%, compared to only 20% for late stage cancer.
Daniel Scrivner (00:11:01):
I want to talk for a second about MRI technology, because I think just to try to describe it for people, I'm going to try to string together a couple of ideas. Let me we kind of dive into this. But one of the things I thought was so fascinating is, I want to talk about why MRI was the right tool. And you talked about specificity, and one other [inaudible 00:11:19].
Emi Gal (00:11:18):
Daniel Scrivner (00:11:19):
Yeah. Sensitivity. That's right.
Emi Gal (00:11:20):
Daniel Scrivner (00:11:21):
But the other thing I think is really interesting is, one, this is an existing technology, and you guys aren't making hardware. So you're effectively saying, "We're going to take advantage of the technology, but we're not going to focus on manufacturing the hardware. We're not going to become experts at making MRI machines. We're going to become experts at basically layering this technology with really interesting software, where we can effectively supercharge the technology." It's a very interesting, very provocative idea. How did you land on that? And I guess talk about why that model makes sense.
Emi Gal (00:11:51):
Yeah. So I landed on MRI, first and foremost because it is the only high resolution medical imaging modality that does not expose you to ionizing radiation.
Daniel Scrivner (00:12:04):
So you do it often.
Emi Gal (00:12:04):
So you can do it every year, you can do it every day for the rest of your life, and you'll be fine, unlike a low-dose CT, or an X-ray, or a mammo where you expose yourself to degrees of radiation, not to say that they're not good modalities, because if you're at high risk for breast cancer, you should get your mammogram, despite the marginal exposure to radiation that you'll undergo. So the main reason I chose MRI was because it is completely safe to do for individuals. The second reason is it has this very high sensitivity and specificity.
For those who are listening who don't know sensitivity and specificity. Sensitivity is how sensitive is the test. If I have cancer and I get tested, how likely am I to find it? How sensitive is it? Specificity is if I find something, how likely is it to be the disease that I'm looking for, in our case, cancer. And so, when you're looking at a screening test, you're looking at sensitivity and specificity, and there's always a trade off between the two, because the higher sensitivity, the lower the specificity, and vice versa. Think of it as a car alarm. The more sensitive your alarm, the more it'll go off, but that means that it will also go off when there's noise on the street, not someone trying to steal your car. And so, MRI has incredible sensitivity. We're talking 96-97% sensitivity, if you look at kind of third-party data. It has an average specificity, let's say about 80%, which means it has an inherent 20% false positive rate.
And our stance is that, while incidental findings can be an issue with screening, if you manage them well, actually the only reason why some people survive pancreatic cancer is because they found it incidentally. And so, what we're trying to do at Ezra is find all of these incidentals and determine which ones are actionable, and which ones aren't. And we built a whole suite of tools to do that.
Coming back to your original question, we have layered a lot of software to make what we do more accessible. Before Ezra, you could get a full body MRI, but it would take two to three hours, it would cost 10,000. With Ezra, it takes 60 minutes, it costs $2,000, and we are working on making it even more affordable in the very near future.
Daniel Scrivner (00:14:33):
I mean, that's already a massive, massive, massive improvement. You've taken the time down by two thirds, you've taken the amount of money down by four fifths, 80%. It's a massive change. One of the things I want to now dive into, and try to structure this conversation around a couple of big ideas, and obviously one that we talked about before, that is very clearly present in what you're trying to do at Ezra, is take something that, as we've described, this is typically most cancer screening today is reactive. So you have symptoms, there's something bad that's going on, and now, I don't know, a bad analogy might be your car's now making noise, and you're like, "Something is not right. So now let me take it into the shop." Well, of course they're going to find something, it may not be great, maybe kind of at the catastrophic phase, and you're trying to change something that's reactive into something that's proactive.
And one of the things that we've had a lot of conversations like this, Andrew Herr at Fount is one of them, we have had the team on from Levels. And there's a big change going on right now just in terms of healthcare, where it is changing from reactive and all the powers held by doctors to proactive, and you as the consumer now are empowered to, if you so decide, go and decide to do this. And so, there's something very powerful in what you're doing of just saying, "Hey, cancer screening, or just looking at your body and seeing how you're doing, you have every right to do it. If you have the money and you're willing to spend it, come and do that at Ezra." Talk a little bit about that, and why that's an important shift.
Emi Gal (00:15:58):
Absolutely. I'm a big fan of... There's this book that was written by Eric Topol, it's called The Patient Will See You Now, I think Dr. Topol Was one of the first doctors to propose the future of healthcare being one in which it is patient driven as opposed to health system, doctor driven. The way I think about it from the perspective of Ezra and what we do, we have the cure for cancer already, it's early detection. That's it. We know how to treat cancer really, really well if it's early stage. The survival rates in some instances are like 90%, 95%, 99%, only if you find it early. The cure is there, we just need to find it early.
To find it early, we need to be proactive. To be proactive, we need tools that are affordable, fast, accurate, and so on. I'm a huge believer in this whole new trend of companies such as Ezra, such as Found, and so on, that are trying to prevent issues from happening before they happen. In the case of Found, they really focus on enhancing performance and making sure that you don't get to even have disease. But if you get to have disease, what we focus on is detected early, because that's when you're most likely to have a really strong chance at remission or becoming cancer free or disease free.
Daniel Scrivner (00:17:32):
Yeah, I mean this idea, we talked about it before this interview, this idea that the cure is already here and it's early detection, I think seems diametrically opposed to how most people are thinking about cancer, which is cancer is this thing that's going to be cured by healthcare companies and at some point in time there's going to be drugs and certain things you can take. It's like, "No, it's not actually real. It's a biological process." You just need to find it early and to find it early, you have to take down the friction and take down the barriers. I love that idea. It just seems incredibly, almost criminally under-discussed.
Emi Gal (00:18:01):
It is insanely under-discussed and actually a big chunk of the medical community is still going like, "Oh yeah, you don't want to look at things because you're always going to find things." The point we're trying to make is that is fine. Let's find all of the things and then do a really, really good job in only following up on the things that are clinically significant. I think a lot of doctors are concerned about, "Oh, we're just going to biopsy so many people with so many things." But the thing is, most of these things that we find never get biopsied. They get tracked over time or you mostly do the follow-up diagnostic scan and then you just monitor things and see how they evolve. I think there needs to be a paradigm shift in how we think about disease and cancer towards early detection.
With cancer especially, the unfortunate reality is you can be a vegan triathlete and still get cancer, because it's a genetic disease. There are lifestyle factors that can impact it, but it can be just like you're unlucky. You have a series of genetic mutations that lead to an oncogene, that lead to a group of cells, that lead to a tumor, that lead to organ being filled with cancer, that lead to expansion, then metastatic disease. What you want is to give yourself the best defense and the best defense is a look inside your body.
Daniel Scrivner (00:19:34):
Yeah. I mean, you hinted on it there, but one of the big things obviously that we talked about before is if you get a scan and it is stuff that is not significant, you also then at least have a baseline to which you can compare, so that the next time you get a scan, you can actually look at deltas as opposed to, again... But this all requires that you're proactive, that it's something you do somewhat routinely. So, it does require this shift.
I want to talk about one more thing. So we talked about, and I'm very excited to check out this book, The Patient Will See You Now, because it's exactly about this obviously massive power dynamic shift that I think has to happen in healthcare in order for outcomes to really change. But the other piece of this too is just ownership of your own health.
It seems like we're going something, we're going through a massive generational shift, I would almost think about it, as I look to the doctor as the person who's in charge of my health too, no, I'm in charge of my health. You just look now at all the wearables, all the amount of supplements, all the amount of biohacking, even just look at stuff like Eight Sleep and Oura Ring and Apple Watch, you already have much more data around your health that's not resided in held, private, inside your HIPAA files that all your healthcare providers have. So, you manage it.
One of the things that I wanted to talk about, just go a little bit deeper on, is, and I brought this up to you when we were talking about this interview is... I started digging into obviously Ezra and one of the things that honestly shocked me a little bit was you sharing on social your experience of going into Ezra and getting a body scan and literally having this massive backlash of basically doctors coming and saying, "Oh, this is terrible."
The commentary that they had around it was, "This is an expensive machine." Almost like, "You and no one else should be allowed to use this expensive machine unless a doctor says so." Around control. The other one was, it's prohibitively expensive, obviously what you're trying to do then is drive down the cost and so it's only this good, that's only some people can have. The second one was around these incidental findings, but I thought it was just staggering the amount of backlash where they weren't acknowledging any of the positives and were just completely focused on the negative. What was that experience like for you? And is that something you come up a time and time and time again with just doctors in the medical establishment?
Emi Gal (00:21:43):
Yes. It's a great point. It's a great question. When I started Ezra, I used to come across that in four out of five doctors that I spoke with. A year and a half, two years in, it was like three out of five. Now, I want to say it's maybe two out of five, maybe even one out of five doctors that we speak with have significant kind of allergic reaction to it. I think what it is, is doctors want to be on the cautious side, for good reason. It's like they are there to, first, do no harm, the Hippocratic Oath. Their concern is legitimate in that, " Hey, if I'm going to do this scan and biopsy a shit ton of people, this is a net negative thing for society," which is true. But the thing that they didn't appreciate is actually you find a bunch of stuff, you don't need to biopsy any of it, and then you just follow up with a diagnostic scan or track the thing over time and see how it evolves. Only after a positive diagnostic scan, do you then go and biopsy people. A lot of doctors that work with Ezra have seen that now over the two, three years that we've been live and have become, from being reluctant to getting a scan, they've become proactive about sending people in. I actually have a very specific example, won't name name, but we have a doctor who was skeptical, who then we did this research project and he decided he's going to try it out himself and scan his family. We found a brain tumor in his, I want to say, 17-year-old daughter. It turned out to be a significant thing. She had it removed and she's now perfectly fine. He became, you can imagine, a big proponent of it, because the probability of his daughter having a brain tumor at that age was just like any doctor would've said, "Oh, probability is so small you don't need to do a scan."
So, we've seen doctors embracing it more and I think, as with any new technology, it's going to take time for the early adopters and then the early majority, late majority and so on to adopt it. I'm playing the long game and definitely working on publishing data and working on partnering with doctors and working on showing that this thing helps people find cancer early and survive.
Daniel Scrivner (00:24:27):
I want to talk about that particular example, because in my mind one of the things coming into this interview that I don't even know where I got this idea was that, okay, so if we are changing from this world where MRIs are reactive and [inaudible 00:24:41] proactive and it's something that you want to do, well, I think about myself, again, I'm 36 and I'm like, "Okay, probably makes sense at my age." But that's a fascinating example where you have a 17 year old who, I don't know if anyone would've said, "Oh yeah, we should get people as young as 17 to go inside an MRI and go and get scanned." I guess one of the questions is just-
Emi Gal (00:25:01):
And we only scan adults, by the way, that was a research project. It just happened, but which was like it was fortunate for that [inaudible 00:25:07].
Daniel Scrivner (00:25:06):
Incredible outcome in that case.
Emi Gal (00:25:09):
Daniel Scrivner (00:25:09):
One of the questions that I just wanted to ask just super generally is what are the guidelines or what do you encourage in terms of how people think about scanning? Because it may be, obviously, just in that example alone, it's like, I'm probably completely wrong to think that you wait until later on in life to do scanning. How do you think about it? How do you encourage patients to think about it?
Emi Gal (00:25:27):
Yeah. We believe it should be the consumers prerogative whether they should get scanned or not with an MRI, because it's a safe modality. If you are 20 and you want to get a scan, you should be able to get a scan. We've certainly have had many people, who were young and not at high risk, who got scanned. The interesting thing is when you speak with doctors, they go like, "Oh, well, under 40, the probability of cancer is so low that it doesn't make sense to get screened." But yet, 150,000 people in the United States every year are under 40 and they find cancer, symptomatically, which means their cancer had existed for a while.
Tell that to the guy who's 28 and just found out he had pancreatic cancer and that they shouldn't get scanned. You can't. We think that what is best for the individual is best for the individual and the individual should have the freedom to choose. The individual didn't have the freedom to choose until recently. I'm actually excited that Ezra's one of the companies that is able to provide that freedom to the people.
Daniel Scrivner (00:26:44):
Well, yeah, and it's not just democratizing access, it's doing it in a way that's bringing down the cost and bringing down the time requirement, which is truly what needs to happen. Because, again, I think just to connect it to a larger picture, forget what conversation that this was, but I've learned a lot about healthcare and some of the statistics just around annual inflation, I think this was from Adrian Aoun from Forward was talking about that I think today, even, if you look at the healthcare costs in the US, it's something staggering. It's like 30% of GDP is spent just on healthcare.
Emi Gal (00:27:14):
18% of GDP.
Daniel Scrivner (00:27:15):
18% of GDP. I think it's headed to 30%.
Emi Gal (00:27:17):
Daniel Scrivner (00:27:18):
But one of the other most surprising things is, and this is not shocking, so if we think that that's growing, it's actually there's inflation embedded in healthcare costs and it's rising at 8% or 9% year over year. We're talking about that now at an economy-wide level and shocked that inflation is 8% or 9%, yet here it is in healthcare and it's been happening for decades now. You guys are truly bringing deflationary forces into at least a small area of healthcare where we need that literally across the board, across the entire service [inaudible 00:27:47].
Emi Gal (00:27:46):
100%. Here's a staggering statistic, Daniel, 80% of the 18% of GDP, so 80% of healthcare costs are attributed to managing end-of-life chronic care, in the last three to six months of one's life. We're spending 80% of our budgets in extending life by three to six months. We should be spending 80% of our budgets in preventing or detecting early, so that you extend life by five years, 10 years, 15 years, not three to six months.
Daniel Scrivner (00:28:23):
Yeah, no, it's very well said. I mean that's a shocking stat.
Emi Gal (00:28:28):
When I found out, I was shocked.
Daniel Scrivner (00:28:29):
Well, yeah, and you combine that with other things. One of the data points that I uncovered in another conversation recently was that right now in the US, I think, and I think this data sum it all, I think it's around 2016, there's around 72,000 people that were centenarians and that's expected to be over a million by the year 2050. You think that if today 80% of healthcare and 18% of GDP is because of end of life, that's going to absolutely explode when you have this order of magnitude explosion in that population.
I want to talk about MRIs, but I want to talk about the patient experience. I'm someone who I have never gotten an MRI, you've talked about before, it's a noisy, I imagine a lot of people probably have a general apprehension or just because they've never done it before, they don't even know what it's like. I'd love it if you could talk through what is it like to go and get an MRI? How long does it take? What do you have to do? What are the steps? Just walk people through that to give everyone a sense.
Emi Gal (00:29:21):
Yeah. The Ezra works is we're direct to consumer, so you go on ezra.com, you sign up for the type of scan you'd like to get and we have various different types of scan with our full-body MRI being the staple product. You choose where you want to get scanned. We have about 17 partner imaging facilities that are hand-curated imaging facilities that are very high quality in locations where we scan, we're live in six cities. You then visit this facility and you get a scan, it's a 60-minute scan, it is longer than you'd want to be in a scanner. It really is. If you're claustrophobic, especially, it's not a pleasant experience. I will say, we've had members now who've done it multiple times in subsequent years.
Once you've done the first one, and you see what it's all about, and you get used to the length of time you're in there and so on, it gets easier. But the first one is certainly hard, especially if you've never done an MRI before. In my case, I've now done dozens of MRIs. I was actually the guinea pig for our very first scan, which was three and a half hours when we were working on this. I find it meditative, because you have no phone, you have nothing, you're just in that machine. The easiest thing you can do is just breathe and meditate.
You get your scan and then five to seven business days from having gotten your scan, you receive your Ezra report, which is a radiology report that walks you through every single finding and a translation of what the finding means. So, we don't tell you just unremarkable parenchyma, I don't know what that means, we translate that for you. Then you can have a video call with a medical provider who's an employee of Ezra, who will guide you through your findings, give you an action plan and next steps, if there are any.
There will be next steps, even if there's no cancer. We sometimes find non-alcoholic fatty liver disease that can lead to cirrhosis, which can lead to other types of issues including potentially cancer. You can actually reduce your fatty liver content by doing exercise, lifting weights, et cetera. So, we tell you what you should be doing based on your Ezra report and then you can have a follow-up call with an Ezra medical provider six months in just to check in and see how things are going. Then, especially over 40, we do recommend that people do it annually.
Daniel Scrivner (00:31:57):
The fatty liver is a really interesting example, that you're not... So you're screening for cancer, you end up finding a fatty liver and then you can obviously help that person, give them advice and coaching around what to do there. Does that happen for in, this is probably a weird question, and I have no idea what this is, which is why I'm asking. Do you find that for other organs? Do you ever find something to do with the lungs that may suggest someone doesn't have enough lung capacity or something to do with the heart? What do you find that you're not looking for besides fatty liver?
Emi Gal (00:32:27):
Yeah, so we find all sorts of things that are not cancer that might be clinically significant. So we often find disc herniations because our scan includes the spine. And so we find disc herniations, which if they're asymptomatic, you don't need to do anything about it. But if your lower back has been hurting and you can barely stand for five hours and so on, now you know why. They have a disc herniation that maybe needs some physical therapy, some exercise, et cetera. We find aneurysms. Aneurysms are important to know about because they can pop. And if they do, you want to know why. We find hernias. Hernias, again, if they're asymptomatic, they're fine, but if they're symptomatic, you now know why your abdominal area has been hurting or whatever.
So we have tens if not hundreds of types of findings. And what we've done is we've created this AI that takes every single one of these findings, converts it into a translation of what it means, and what you should do about it, and then we put that into the Ezra report that you receive. So we're very focused on cancer, but you get significantly more value than just screening for cancer from doing an Ezra scan.
Daniel Scrivner (00:33:41):
I want to go a little bit deeper into AI. And part of this is... I guess one of the questions I wanted to ask is, so you guys have managed to take an existing technology, MRI technology, and significantly drive down costs and significantly drive down time. And I imagine maybe part of that is because you're very specifically looking for very specific things. You're checking for, I think now, tens of different types of cancers that you guys can spot. Am I getting that right or is the AI that you guys have designed, would it be able to broadly bring down the cost and the time required for an MRI? And I guess what I'm trying to get at there is, was AI the right solution here because you're looking for something very specific, and is it a bad general purpose way to try to drive down time and cost?
Emi Gal (00:34:27):
Yeah, so the answer is both. Our first effort in decreasing scan time is that we designed a recipe for scanning that was tailored for screening. And that by itself helped us reduce scan time from three hours to one and a half hours. And then we layered in AI to accelerate the scan time even further. And then we layered in AI to accelerate the interpretation time even further, and so on. So our stance with machine learning is that it can be used to automate things or to enhance things. And we don't think ourselves as an AI company, we think ourselves as a cancer screening company that leverages AI to decrease the cost of cancer screening. And can't talk about too much, but next year we have some stuff launching that will significantly decrease the cost even further from the current levels, which are already pretty good, from 10,000 to 2000.
Daniel Scrivner (00:35:36):
What are the other questions I wanted to ask is, probably I, like a lot of people listening, have heard a lot about AI and yet, I at least for myself, haven't worked at a company that has... As you guys say, yes, developing AI is not the mission of the company. You want to make it much, much, much easier, much more affordable, much faster for people to screen for cancer. But AI seems to be the right tool. And so one of the questions I wanted to ask you is, what has it taken for you guys to be world class at AI? And I guess what I'm curious there is everything from, what does the team look like, to how is this different? How is building AI software maybe different than software you built previously?
Emi Gal (00:36:17):
So AI, especially in healthcare, boils down to one thing as being the most important thing, which is data. The neural net that we use at Ezra is not state of the art reinforcement learning, which is the most popular stuff you see these days in DALL-E and stable diffusion and so on. But the value of what we've built at Ezra is all in the data. We have what we think is the largest data set of full body MRI images out there. We have created internal processes for curating data sets for training AIs. We have internal teams that are responsible with running what's called verification and validation for AIs. When you create an AI and you want to put it in production, you need to have it cleared by the FDA. To do that, you need to validate that the AI works on your intended population, on the types of magnets you want to scan on, et cetera.
So we have an incredible team at Ezra and partnerships that allow us to acquire all of this data that we need in order to validate AIs. So I think in building AIs in healthcare, the data component is the limiting factor and we have a huge advantage at Ezra because we have a lot of data and we have a lot of process internally on how to acquire this data. So from a team structure perspective, cause you asked about that, we have an incredible scientific officer who leads, especially our MR team, who is the inventor of parallel imaging in MRI, that's Dan Sodickson, and who's the Vice Chair of Radiology at NYU. I think he's now Chief Scientific Officer at NYU, that's his title. And then he also spent part-time with Ezra.
We have a team of AI engineers, we have a team of AI infrastructure, we have a team of regulatory affairs, and we have a team of operations. And all of these teams work together to acquire the data, to train AIs, iterate on AIs, acquire the data to validate, validate, write 900 pages, submit to the FDA, get clearance, and then move on to the next one. So the biggest surprise for me in building AIs for healthcare is how much time is actually spent on data, moreso than the NeuroNets, themselves.
Daniel Scrivner (00:38:52):
Yeah. I imagine obviously, having to get FDA approval and having the regulatory aspect is obviously very unique as well too. You alluded to it, I'm sure you're being completely serious, that you do need a 900 page report that's going to the FDA.
Emi Gal (00:39:05):
Exactly. A report that you submit to the FDA for clearance is roughly 900 pages.
Daniel Scrivner (00:39:11):
So if you were to try to estimate, and not that this can go away, and obviously that's a bad thing, that if you're building technology and AIs in healthcare, that it needs to go for validation and testing to some external third party. That's not a bad thing. I'm not saying it is. But if you were to think about the pie chart of how much time you guys spend on AI, what percentage of that is dedicated to just managing the regulatory aspect?
Emi Gal (00:39:32):
Frankly, not a ton. I've actually been very positively impressed by the FDA and the quality of people at the FDA. They really know what they're doing. And having a body that's external, that's unbiased, that's focused on protecting consumers with safe and effective devices, creates a really useful forcing function for companies because you really need to get all of your quality management in place, processes in place, validation in place, and so on. And so having the FDA as the regulatory body that we need to submit to whenever we create one of these Ais, is actually internally for us, really, really useful because it forces us to get all our ducks in a row.
And so we would have to spend all of this time to run V&V and so on, regardless of whether we filed with the FDA or not. It so happens that filing with the FDA creates a really clear business reason to invest in all of those processes. So I know it sounds odd, but I've actually quite enjoyed the process of working with a government body like the FDA, because I can see the value of putting out there a device that's safe and effective.
Daniel Scrivner (00:40:52):
Well, and as you said, it raises the bar for rigor internally with which you're doing these testing, with which you're distilling down and making decisions and deciding what to do and what not to do.
Emi Gal (00:41:04):
Daniel Scrivner (00:41:06):
I want to talk about one more aspect of your business, and I'm curious whether it was intentional or not. I imagine it was. But one of the things that's fascinating when I think about Ezra is, so you're focused on cancer screening. The main thing that you're investing in is the software, the AI, the technology that you're using to basically process and detect cancer and actually be able to figure out what to do. And so by nature of that, you have what a lot of people would call an asset-light business model, just meaning, you're not building the real estate from the ground up to have these scanning offices, you don't own the MRI machines, and you're able to take advantage of centers that already have them.
So I guess what I'm curious is, was that intentional from day one and how does that work? I'm guessing it's because there are dedicated service centers that just do imaging and so you can say, "Hey, you guys have the MRIs. We can bring the technology and the patients." Talk a little bit about that, cause I think it's fascinating.
Emi Gal (00:42:02):
Yeah, it was a very deliberate decision. I looked at the space and realized that medical imaging is a really competitive, very commoditized, low margin business. In other words, it's not the type of business you want to allocate a lot of CapEx into and get into if you're not already an established player. In looking at the space, what we also realize is a lot of imaging facilities have remnant inventory. They are not operating at 100% capacity and therefore we saw clear opportunity to leverage that. And so the way Ezra works is we partner with the best imaging networks in the country. Our largest partner is RadNet, which is a public company, the largest outpatient medical imaging network in the country. And we buy MRI scanning time from them. We run our protocol and our software and our radiology templates when you go in and get scanned and then we get all of the images, the reports, and then deliver all of the consumer experience.
One thing that is important to note is that in building a lot of software, we didn't actually just focus on the AI part. We've essentially built an end-to-end cancer screening stack. The booking system, the integration with the facilities, we built our own EHR, we built our own report generation system, et cetera, because the current healthcare system is not optimized for a great customer experience. Even booking a mammogram, which should be the easiest thing to do, and a woman has to call a facility, stay in line, all of that stuff. With Ezra, booking a full body scan, it's as easy as booking an Uber. It's not easier, it's three taps and you're done. And so we believe in increasing access and broadening the ability for people to get a scan, not just through making scans faster and creating full bodies and so on, but also by applying our skills to the underlying technology to actually book and manage and reschedule and do all of that stuff. And we're really proud of what we [inaudible 00:44:23].
Daniel Scrivner (00:44:23):
Well, and it's an important point because all of those things are major inputs into the customer experience, how customers feel going into the appointments, how happy they are having chose Ezra. And again, I think it's another just fascinating example of, I imagine you guys have similar views, but just trying to do anything medically, it just feels like you're in the stone ages if you work in technology and use Calendly and use Ubers. You're like, "Why can't it be this simple?"
Emi Gal (00:44:48):
Absolutely. And here's a crazy stat, there're about 60, six zero, million women in the United States who should get a mammogram every year and only about 40 million are compliant. So you have 20 million women who are uncompliant with their mammogram screening. I have to believe that part of that is because it's just hard and they don't have time to handle the logistics of doing that.
Daniel Scrivner (00:45:15):
We're humans. It's like it's friction. You don't want...
Emi Gal (00:45:18):
It should be frictionless and it should be super easy and you should have all your data in one place. And we need to do better as a health system to get there. And there are many things we're working on at Ezra in order to make all of that stuff better.
Daniel Scrivner (00:45:34):
Well, and just to say it even more starkly, if you could choose between having that experience to book a car, which is not that high value, and having that experience and the lack of friction to go and do something that you should absolutely be doing for your own health and wellness. This should exist in healthcare first and foremost. But unfortunately it doesn't. To take it over from consumer.
Emi Gal (00:45:54):
It's not,, but I think we're seeing the early signs of these things being resolved through investments by startups predominantly.
Daniel Scrivner (00:46:02):
I want to ask one big zoom out question and then I want to end by talking about a few lessons learned. And the question that I want to ask is, Ezra, to me... We've talked about a bunch of things here, of going from reactive to proactive healthcare, of people having ownership, of patients being in charge of their healthcare, you mentioned that book, The Patient Will See You Now, even this idea that you guys are bringing truly deflationary forces with technology into something that is inherently inflationary, that's catastrophically inflationary now. And so to me, Ezra seems like this fascinating, very optimistic, and exciting glimpse of how healthcare should be. And it's something that maybe feels like it's a decade out, probably decades out, from rippling out across other areas. I'd be curious for your perspective, if you try to jump forward 10 years, 20 years in time, what do you hope that healthcare starts to feel like? I imagine you probably have some pretty specific ideas.
Emi Gal (00:46:53):
Absolutely. Yeah. So I started Ezra with a very clear sense of mission. And our mission is to detect cancer early for everyone in the world. And we have this very clear North Star, which is every year, about 10 million people find cancer late. And so our mission at Ezra is in a decade to 20 years from now, and I personally plan to work on this for the rest of my career, we are finding cancer early for those 10 million people, which means we're screening probably 100 million people because we're not going to find cancer in every person we screen. And so if we find cancer in 10% of our members, we'll have to screen 10 times more members in order to find cancer early for those 10 million people. So that's Ezra.
If I zoom out and I look at the healthcare system and healthcare in general, and try to draw a line to what I think might be true a decade from now. Cause I think a much more significant part of healthcare is about preventative care, about proactive, early detection, about holistic kind of approach to healthcare as opposed to treating people who are already in a chronic disease state.
Daniel Scrivner (00:48:16):
So moving from reactive to proactive and effectively just that rippling across everything. Well, and this is something that's come up before where it's like healthcare, I'm sorry, but it's a misnomer today. It's like what it actually is sick care today. And what you actually want to move to is actual healthcare where you are caring for your health when you're healthy and it all revolves around proactivity.
Emi Gal (00:48:38):
And here's the crazy thing, it's actually a lot simpler to invest in preventative and proactive care than it is to treat people who are in kind of various degrees of chronic disease. It's much easier to prevent diabetes than to treat diabetes. It's much easier to do a partial surgery for removing a malignant small breast tumor than to do mastectomy plus radio plus chemo. So it is investing in preventative care and investing in detection just makes logical sense for a rational individual. And the tools are there, but they're not kind of evenly distributed just yet.
And I think we need to invest.
Daniel Scrivner (00:49:32):
Totally. I mean, this may be a silly tangent, but it almost seems like what if we applied healthcare as it exists today to software development. And you said we're not going to focus on great coding and doing all this testing up front. We're going to be kind of lazy there and we're just going to invest all our resources in handling catastrophic issues when the software fails and when the software goes down, everyone would be like, what are you doing? You need to work way more upstream. Are you taking that approach?
Emi Gal (00:49:58):
100%. And I feel that somehow unfortunately the incentives are misaligned in healthcare, skewing towards treating sick patients instead of preventing disease. But with value-based care and everything in that kind of realm, I think things are getting better on that front.
Daniel Scrivner (00:50:18):
Yeah. I think as well just generationally, you also now have, I think of myself and anyone younger, I think they all inherently see and think of their health as something that they need to manage. It's on them. They're tracking it. So it's a very different perspective.
Emi Gal (00:50:36):
Preaching to the choir. Yeah. Yeah.
Daniel Scrivner (00:50:36):
Okay. So I want to end just with a couple of points talking about what it's been like and some of the lessons you've learned building Ezra. And one of the questions I always want to ask for founders listening to this, everyone just building a company is it's an emotional exercise. You have super high highs, you have super low lows. So I'm curious for you, what high and what low comes to mind that you've had on the experience building Ezra so far that you'd be open to sharing?
Emi Gal (00:51:02):
Yeah. So the biggest tie was when we found cancer for the first patient. And that patient reached out to us after having gotten treated and he was like, "You saved my life." Actually, I remember the email that he sent me an email directly and the title of the email was Ezra Saved My Life. And then he wrote a heartfelt long email about how we saved his life. And that was the peak of my Ezra experience was that very moment. The lowest of the low was when the pandemic hit and we went from scanning hundreds of people a month to scanning zero people a month because just facilities were shut down, everything was shut down and it was hard for the business. But it was also hard because I knew kind of intuitively because of what I do that a lot of people will find cancer late because we're going to have a period of 12, 18 months where nobody's going to get screened or very few people are going to get screened. And as it turns out, about four or five months into the pandemic, we started scanning people again and everything was fine. But I've started seeing data from the UK, for example, they published a big study showing that late stage breast cancer increased by 20%, like an astronomical number because women skipped a mammo cycle.
So, that moment during the pandemic was the lowest. I will say though there's that quote from Elon Musk that I always come back to, which is building a startup is staring chewing glass while staring into the abyss. That's how I feel most days. It's kind of like I love what we do at Ezra. We're building important things. It's great work and we find cancer for folks. But it's hard. It's like building a startup, especially building a startup in healthcare is incredibly hard. And so those weekly reminders when we do something good for someone are useful to help us keep going.
But most of the work that goes into building a startup is just hard work that requires a lot of discipline. And I think not enough founders talk about that.
Daniel Scrivner (00:53:35):
Yeah. No. It's like the outcomes. Finding cancer for people, which is the whole mission of the company, is incredibly gratifying, but the work to get there is incredibly dull. You have to just do that every day.
Emi Gal (00:53:46):
Do that every day and wake up and set the goals and just pursue them. And it's especially hard when you're doing something incredibly new in a really difficult space. We're trying to literally find cancer for folks with a new technology.
Daniel Scrivner (00:54:01):
No, I have a lot of empathy for what you're going through.
Emi Gal (00:54:05):
Daniel Scrivner (00:54:05):
I want to ask two more questions and one of them is, so Ezra is the second company you've built. The first company is Brainient, which we didn't even really cover much except for the intro at the beginning. But one of the questions I want to ask, obviously anybody who is building their second company I imagine has things they've learned or ideas about what they want to do differently. Does anything come to mind and was there anything when you were building Ezra that you were very clearly I'm going to do different the second time around?
Emi Gal (00:54:32):
Yes. So the nice thing about being a second time founder is it's easier to raise money because you have the credibility, you've made money for investors, investors trust you. That part was significantly easier in Ezra than it was in Brainient. The learnings part I learned, and we've done a great job with that at Ezra, that it's really important to have a strong sense of mission. With Brainient, the kind of mission was I got to make some money because otherwise I have no money. And that was the goal. So there was no kind of stronger sense of mission that would rally the troops and get everyone excited.
With Ezra, we do have this mission, we all care about it. It's really important. So I think as far as some insight for founders is because startups are about chewing glass while staring into the abyss, you really want to care about the thing that's getting you to chew so much glass. Because if you don't care about the thing, it's not worth it. There are many other ways to make money that require less glass chewing. So I think it's really important to have a strong thing that you care about, that you want to invest in, that you are willing to go through great depths of pain to pursue.
I think that's my ... Doesn't sound like a very exciting thing, but that's my biggest insight and, I guess, piece of advice that I have for founders. Choose something that they really, really care about.
Daniel Scrivner (00:56:11):
Yeah. And then thread that needle and build a mission driven company and hire people for that and have that be the rally and cry.
Emi Gal (00:56:17):
Yes. Exactly. That really makes it easier to find the right team and get the team rallied towards the short term goals.
Daniel Scrivner (00:56:28):
Yeah. I would consider Ezra to be incredibly successful. And I know you guys are still early in the existence of the company, as you said, you want to work on this problem for the rest of your life. One of the questions that I find fascinating and I like to ponder about different businesses, and so I'd love to ask you, is if you had to point to one decision, one reason why Ezra has been so successful, what would you put your finger on? Meaning, if you had to try to boil down just the trajectory of the company into one thing, what would that be?
Emi Gal (00:56:58):
Yeah. So it's quite counterintuitive, but I think our biggest strength has been the fact that I am an outsider to the healthcare system because any insider would've come and said and come with a set of preconceptions that would've prevented them from pursuing things in the same way I did. I went with a direct to consumer approach with a new way to scan people with a modality that had not really been used before for screening and with a new kind of type of business model for serving my goal, but I guess a little bit naive about all of the different things.
And that was a good thing at the end of the day. So I think the fact that I pursued a domain that was where I didn't start with insider info served us really well. And actually, I think that's a lot of times you see founders going into very new spaces and succeeding because of the fresh eyes on the problem.
Daniel Scrivner (00:58:10):
Well, and I think the way you broke it down there, of all the decisions that you made that were very counter to how somebody would've made those decisions inside, I think totally proved the point because those are one-way doors. You're going to make those decisions once and it's either going to work or it's not, and you happen to make a very different set.
Emi Gal (00:58:28):
Daniel Scrivner (00:58:29):
This has been so much fun. Thank you so much for coming on, Emi. And I highly encourage if you have any closing remarks, feel free to share them. But I highly encourage everyone to go and check out Ezra and consider getting a screening if they hadn't had one already.
Emi Gal (00:58:44):
I think that if I am to have an ending remark, it would be to not smoke, exercise and get screened.
Daniel Scrivner (00:58:52):
The three things. The three things to do coming out of this interview.
Emi Gal (00:58:56):
The three things to do at the end of this interview. Daniel, this is super fun. Thank you so much.
Daniel Scrivner (00:58:58):
Thank you so much for coming on, Emi. I appreciate it.
Emi Gal (00:59:00):
Daniel Scrivner (00:59:01):
Thank you so much for listening. You can learn more about Ezra and how to get screened for cancer at ezra.com and you can follow Emi Gal on Twitter @emigal. That's at E-M-I-G-A-L. You can find a searchable transcript for this episode as well as our episode guide with ways to dive deeper into the topics we've covered today at outlieracademy.com/146. That's outlieracademy.com/146. For more from Outlier Academy, follow us on Twitter, LinkedIn, Instagram and TikTok. Subscribe to our free weekly newsletter at cheatsheetnewsletter.com.
Subscribe to our YouTube channel where we have all of our interviews that we've ever released, as well as our favorite clips from every episode at youtube.com/outlieracademy or visit outlieracademy.com for more incredible Outlier Founder episodes profiling incredible companies like Forward, Eight Sleep, Commonstock, Varda Space Industry, Superhuman Primal Kitchen, 1-800-GOT-JUNK and many, many more. In every interview, we deconstruct the ideas, frameworks, and strategies they use to build these incredible companies. We'll see you right here with a brand new episode of Outlier Academy next Wednesday.
On Outlier Academy, Daniel Scrivner explores the tactics, routines, and habits of world-class performers working at the edge—in business, investing, entertainment, and more. In each episode, he decodes what they've mastered and what they've learned along the way. Start learning from the world’s best today.
Daniel Scrivner and Mighty Publishing LLC own the copyright in and to all content in and transcripts of the Outlier Academy podcast, with all rights reserved, including Daniel’s right of publicity.