Algora | Algora TV

Open source interviews #25 - Cody Ebberson, founder of Medplum.com

Algora

Watching now

00:00 Hey Cody, thanks so much for joining me.

00:06 Good, how are you sir?

00:09 Doing great, it's great to meet you, thank you so much.

00:13 So far in this podcast series, the first Open Source founder building for healthcare

00:18 and so I'm super excited to hear about your background and how Medbloom got started.

00:24 Awesome, well thank you for having me and thank you for talking about Medplum. We're

00:28 super excited about it. And Medplum, I've been in, the Medplum team has been in the

00:35 healthcare tech industry for like 15 years now. And I think that's given us an opportunity

00:41 to really work with a lot of different healthcare organizations, understand a lot of the problems that our partners and our customers face.

00:49 Healthcare in the United States is a very complicated space, lots of regulatory requirements

00:54 and security requirements, interoperability requirements.

00:58 And I think that's given us a good kind of bird's eye view of all the different challenges that people

01:05 face. I think that in general, in healthcare or in tech,

01:10 people are very spoiled.

01:11 There's lots of great open source libraries out there to make new

01:14 applications fast and quick and easy.

01:16 And that a lot of times doesn't apply to healthcare. For example,

01:21 you can't go use Vercel with a healthcare application because they're not HIPAA compliant.

01:27 And so it creates an interesting space for us.

01:33 Absolutely. Huge gap in the market.

01:36 And it sounds like you're the right team to be building this platform.

01:39 And so, you know, congratulations for this huge undertaking.

01:43 What did the early days look like?

01:46 Yeah, it's a good question.

01:48 So maybe just a little bit of backstory, which will give some context.

01:53 Our first experience in healthcare was almost 15 years ago, where we were doing something

02:00 called teleradiology.

02:01 And we were based in Asia at the time.

02:03 And at that time, it was kind of more of the wild, wild west when it comes to cloud and healthcare. But the idea was

02:10 in the United States, nighttime radiology is very expensive. And so you could have American

02:16 radiologists living in Asia. So it's daytime there when it's nighttime in the United States.

02:20 You would digitize these radiology exams, send it over the internet.

02:26 They would look at the images,

02:27 make the report and send it back.

02:29 But it's nighttime ER work.

02:31 And so the turnaround time was expected to be 20 to 30 minutes.

02:33 And so if you have a data set

02:34 that's hundreds of megabytes or gigabyte in size,

02:39 you need to transmit it to Asia

02:41 where your internet bandwidth

02:42 might not be the most reliable.

02:44 You need to then view it and then send it all back.

02:46 It was quite a complex process,

02:48 which was good because there's a real trial by fire for healthcare tech and

02:52 healthcare interoperability.

02:53 We've been the same team's kind of gone through a few different products and

02:56 in the radiology space and then in lab testing space and in general

03:01 like general medicine. And so through that process,

03:06 we've accumulated a healthy network,

03:08 which is, I think, always really helpful

03:10 for any startup, right?

03:11 Having lots of connections and people in the space

03:14 who you can lean on for like beta testing

03:18 or a sounding board is critical.

03:20 So then the earliest of days,

03:22 it was mostly just with other healthcare startups

03:24 that were our friends. And that was before we had fully committed to the idea of like, oh, this is a startup. This is a business. At that point, it was more of a, we have friends who are doing some interesting healthcare work. They need a place to store data. We know how to do it securely and compliantly. So we'll just help them out.

04:05 So we'll just help them out. And I think that goes a long way to one, building goodwill with your early customers and your early partners. And two, validating that there is a reproducible need that like, you know, dev shop or a professional services organization, that all you are is just doing contract development work and there's

04:10 nothing that's going to be reproducible instead of the center.

04:13 And so I think that was a big part of the initial first year or so was just validating

04:17 that hypothesis that there was enough of a kind of common platform that we could use

04:22 with all partners.

04:24 I love that.

04:25 And was that open source from day one when working with these teams?

04:29 No, it was not day one.

04:32 The initial days, I mean,

04:34 when you're doing your initial cowboy coding phase of things,

04:39 of like the V0 of a project,

04:42 I think that doing it with the open source polish would be intimidating.

04:46 But I think very early on, the beginning of 2021,

04:50 and it's been open source since that time.

04:53 And which is also somewhat intimidating.

04:57 I think some customers love the fact that it's open source

04:59 and that they have that much visibility into the process.

05:03 And there's others who are scared of it.

05:05 And they're like, wait, does that make it less secure?

05:07 Or can hackers use this against us?

05:11 And so there's been some interesting conversations

05:13 to try to mitigate those concerns over time.

05:16 But no, it's been open source for the past two years

05:20 in the production setting for that time and more,

05:25 which is, and it's going great.

05:27 That's phenomenal. That's phenomenal. And six months ago you had,

05:30 you launched on Hiker News, right? And the reception there was real big.

05:33 Would you like to talk about this?

05:35 Yeah. So we, thank you.

05:37 We went through Y Combinator as part of the Summer 22 batch and

05:42 and then the YSE partners are very adamant.

05:47 You need to get out there and launch.

05:49 And I think that all startups,

05:51 especially software engineers,

05:53 are maybe oftentimes a little bit timid

05:58 about launching on Hacker News.

06:00 It's a very opinionated audience

06:04 and they have lots of thoughts and feelings about

06:07 how things should be done. And so I think that we wanted to really make sure that the house was

06:11 clean, the table was set, that we were ready for a very harsh and critical audience to kind of come

06:17 in and judge our product. But the launch went great. The community was very receptive to it.

06:25 I think that there were a lot of people from healthcare

06:27 who understood the need and the opportunity immediately.

06:30 A lot of people who aren't in healthcare

06:32 who had a kind of like,

06:34 why doesn't this exist already?

06:36 Which is, I think, a very reasonable response.

06:39 So overall, it was great.

06:41 And it's also great from a business development perspective.

06:44 It's a chance to

06:45 um kind of debut to an audience of potential customers and get their eyeballs get their

06:51 feedback and start those conversations that's that's phenomenal and ever since then are there

06:56 some uh you know maybe highlights you might like to share in terms of users, contributors, the community. Yeah. Absolutely. Yeah.

07:05 And in a way, I think that some of our GitHub numbers, I think that people would look at

07:12 it and think, oh, those are rookie numbers, right?

07:14 Like if you compare yourself to something like React with like, what is it, 20,000 stars,

07:20 it's a very high bar.

07:20 But if you just stay focused on the healthcare open source ecosystem, it's been a very high bar. But if you just stay focused on the health care open source ecosystem, it's been a very fast rise.

07:28 And we're now in the top three or so projects in many health care categories, which has been very satisfying and rewarding.

07:37 We have some very significant customers today, including Roe or Roman Health, which is a very large at-home lab testing company.

07:49 30 Madison, which is, they have a number of different brands, including Nurex, which is one

07:54 of the largest online birth control distributors in the United States. And so these are significant

08:01 operations with hundreds of thousands, in some cases, millions of patients. And that they've trusted MedPlum as one of their primary components in their tech stack, it feels very rewarding.

08:13 And those are just the large ones.

08:15 Then there's also a very vibrant ecosystem of early stage companies that are building completely new digital health care experiences, both in the United States and internationally. And in some cases, in many cases, we don't have a commercial relationship

08:29 with them. They're just using the project and we only hear about it because they ask a question

08:33 in discord. And I think that as a team, we are strong believers in the open source ethos. And

08:41 that is awesome. It feels great that the project can start to take on a life of ethos. And that is awesome. You know, it feels great that the project

08:45 can start to take on a life of its own. And even without a commercial relationship that people can

08:51 find value in the project, and it's helping them to do new and innovative things in healthcare,

08:57 it makes us it's so satisfying. I love it. Thank you so much for sharing this. And I think it's

09:03 no exaggeration to say that you're trying to become the standard in the industry for people building healthcare applications, right?

09:11 Yeah.

09:12 That's the plan over here. And how are you managing all this growth in terms of the team and hiring plans, funding that you secured? If you'd like to talk about these things. Yeah, good question.

09:25 So when we started, there was three founders

09:29 who we've known each other for a long time

09:32 and working together for a long time.

09:34 We did raise a seed round in 2020 too, so last year.

09:41 And that's enabled the growth.

09:43 The team is now six.

09:44 We've hired some really talented software engineers,

09:47 people that we've worked with in the past who I think is,

09:50 it's just critical in the early days, you know,

09:53 going to your personal network. So I think anytime you hire an engineer,

09:58 there's always two questions. One, can they actually do the job? And two,

10:02 can you work well with them? Right.

10:03 Cause personal dynamics and interpersonal conflict

10:07 can kill a startup.

10:09 So going to folks that we've worked with in the past

10:11 and we know that they're like top tier engineers

10:14 and great team players was really important to us.

10:17 And thankfully that's happened.

10:19 So the team is now six,

10:20 which has really increased our overall capacity

10:24 to serve our customers.

10:26 I think there's like a spectrum in open source land, right, of things that are, you just kind of give the code away and you hope that people are going to use it.

10:36 And you may have very loose relationships with all your users.

10:39 And then there's the other end, which is more enterprisey, which is where we're at, where when someone engages with us, like we're all in with them and we're going to have weekly

10:49 meetings and shared Slack channels.

10:51 And we're going to do training, onsite training and lots of guidance on architecture and deployment.

10:59 So those big deployments are somewhat heavy lift, but we think it's also very important

11:03 because especially for our first,

11:05 however many customers,

11:06 we want to make sure that they have an amazing experience, right.

11:10 That like, that they are not only did it solve their problem,

11:13 but they loved the process,

11:16 which people don't usually think of with software engineering, right?

11:18 Like, Oh, I love that library, but that's, that's our goal that they,

11:21 it's so delightful to them.

11:24 And they feel like they were able to jump forward in terms of months of unnecessary development that we just cut out and it allows them to launch faster or open up new capabilities.

11:37 And so we think it's really important to work really closely with those initial customers, which is hard, but I think it's very worth it.

11:42 with those initial customers, which is hard,

11:44 but I think it's very worth it.

11:48 And then after that, for the broader ecosystem,

11:50 it's been an interesting education.

11:53 We're trying to follow the kind of best practices playbook of other open source projects with a Discord community

11:56 and trying to engage our users there,

11:59 taking lots of feedback.

12:01 And we don't have crazy numbers in contributors.

12:04 I think we're at 30 today. and every single one of them is very

12:08 satisfying and it's hugely rewarding.

12:12 Sometimes it ranges from, you know,

12:15 cleaning up some docs and sometimes it's, wow,

12:17 they really rolled up their sleeves and they're deep in the weeds on this

12:21 database connection, blah, blah, blah.

12:24 And whenever like someone really goes crazy like that, that's the best because to me,

12:30 it signals the future of what's possible of, you know, more meaningful and impactful

12:35 contributions at a very deep and fundamental level.

12:39 Absolutely.

12:39 And of course, you know, building applications and templates that other people could use

12:43 to fast track their own development.

12:46 Absolutely. To support the sustainability of the project and have the capacity to invest in these relationships with your customers and service them.

12:55 Is there something we could say here about how you approach monetization and other founders, you know, with similar markets in a similar stage, how they could navigate it?

13:05 Yeah. Yeah, it's a really good question.

13:08 So for our largest customers,

13:09 and this is actually quite rare in healthcare technology,

13:15 almost every healthcare tech product today,

13:18 if you go to their website and you have this question in your mind of like,

13:23 how much is this going to cost?

13:25 And almost nobody will tell you, right?

13:27 It's always contact sales, schedule a demo.

13:30 And there's this feeling of like a kind of sleazy process that it's going to be a whole

13:35 bunch of salespeople jumping on a meeting.

13:38 And we want to avoid that.

13:40 So we have publicly available pricing, which for there's like a very basic tier. If you

13:46 are, first of all, people can use the product for free. And if you want to get some basic level

13:52 support, if it starts at $300 per month, the publicly listed pricing goes up to, I think,

13:57 about $2,000 per month. And then we have additional plans that go beyond that. And that really depends on what level of support or custom feature development or, I guess,

14:10 what kind of quirky configuration or deployment process do you have that will require our

14:15 time and energy?

14:16 And it kind of goes up from there.

14:17 And I think that there's an interesting dynamic in the United States healthcare system where early players,

14:28 their overall revenue and willingness to spend money starts relatively low.

14:35 But then if you kind of just follow the line of like, as organizations grow, it stays low.

14:39 And then at the very late stage, the numbers get wildly, outrageously large.

14:45 And like, if you think about, so I live in San Francisco, the largest hospital system here is UCSF,

14:51 the University of California, San Francisco.

14:53 And their primary healthcare platform is Epic, which is the most commonly used platform for

14:59 most hospitals.

15:00 And this is publicly information, but they spent about $700 million on their Epic deployment.

15:06 And that's a lot of money.

15:08 And I don't know over what period of time that applies to, but there are many large

15:13 hospital systems that spend that much money.

15:14 So you have this like long, shallow curve.

15:17 And then at the late stage, people spend massive amounts of money.

15:20 And we're very early.

15:23 We're not eligible for contracts like that. We won't be for a number of years, but that we need to kind of go through this process, eat our vegetables, develop our reputation, build out our capabilities, build out our ecosystem that like over time that we could target opportunities like that is a really exciting idea. And there, there hasn't been, in my opinion,

15:45 there hasn't been enough people challenging that status quo.

15:49 And if you think about it, it makes sense. It's a, it's a gauntlet.

15:52 It's going to take decades to like get to that level, but perhaps naively,

15:56 like we were going to make a run for it. We're going to try.

15:59 And so it's just a matter of like,

16:01 how do we continue to kind of grow and move up the ladder,

16:04 move up the sophistication, move up the size of our clients and customers and capabilities?

16:08 Absolutely. Absolutely. Wow. That's that sounds exactly right.

16:11 That's that's phenomenal. And as the pool from the market and from different niches of the markets of the different size increases, how do you go about, you know, maybe prioritizing which conversations you should have, which customers you should work with more closely. This can be challenging, right? It's very challenging. You're

16:29 absolutely right. And you often have these kind of competing goals. And so we're coming at it from a

16:37 few different angles. And so I'll very transparently kind of share how we think about this.

16:42 The most immediate goals that we have are, you know, we want to have

16:46 healthy revenue and like, you know, a viable business. So revenue is good. And so if there's

16:51 an opportunity that's going to like generate revenue, let's do it. We also are kind of going

16:58 through the venture playbook. And so, you know, we raised a seed round. We have our eyes on what

17:04 are the kind of milestones and metrics that we would want for raising an A and raising a B. And what does that

17:09 like overall curve look like? What does that path look like? And in our mind, there's certain

17:13 milestones and key capabilities that we want to make sure that we have in the platform,

17:19 because they tell the story about that kind of ongoing curve. And so certain examples are getting access

17:28 to certain payer networks. And so being able to take or receive payments from Medicare and Medicaid,

17:33 you know, there's a process and you have to, it's not just a technical limitation. There's a number

17:39 of legal and compliance and bureaucratic challenges that you have to go through. Same with ordering medications or ordering controlled substances or all these various

17:49 like things that we will have to do eventually to tell that story.

17:53 And so then there's now this strategic element when we're talking to potential customers

17:58 of, you know, if they're paying revenue, that's great, but maybe they're not going to be that

18:02 strong on revenue, but they're going to

18:05 be a kind of design partner for getting access to this new capability. And we have like a list of

18:11 10 or so key capabilities, like tent post capabilities that we really want to be able

18:17 to demonstrate. And so we're very happy to work with customers on a strategic basis because,

18:22 you know, we can help them solve their problems and they can help

18:25 sponsor the development of these necessary milestones. And so we have a lot of internal

18:31 conversations just trying to navigate and find balance between those two goals.

18:36 Absolutely. Thank you so much for sharing this. Actually, this makes a lot of sense. And so both

18:41 the roadmap and the relationships and where you focus, they all go hand in hand and you optimize for getting revenue and then satisfying some of those use cases that, you know, need to be prevalent.

18:51 That's that's that's excellent.

18:53 Is there something we could share with other founders who hope to build in the space of health care in terms of the adoption of products in terms of, you know, this is for developers, right?

19:07 But there are big organizations and maybe those developers are not decision makers.

19:11 And so how do you approach this, right?

19:14 It would be amazing.

19:16 Yeah, no, this is a great, great question and a great topic and a very complicated one.

19:22 First, I'll make a few distinctions that I think are important distinctions to make.

19:28 One is that we serve the provider side.

19:32 So in particular, organizations that are going to have their own doctors are working with

19:38 doctors.

19:39 There is a separate category of products which are really going after patients directly and so it's

19:47 usually our customers are going after patients but they have their own doctors or some kind of

19:52 medical provider on staff um going after patients directly is a very difficult business because

19:57 the cost of customer acquisition is very high because you're typically competing with pharma and large consumer brands that have,

20:08 they're just going to outspend you. And it's really difficult unless you're doing something

20:11 very, very clever, very, very innovative. It's very difficult to acquire customers in that space

20:17 versus like working with doctors and working with healthcare providers.

20:22 I think that that's a more manageable and there's just a much

20:26 larger market there. And they have more reasonable expectations. And you can go through more of a

20:33 traditional B2B sales process with those groups. And so I think that that's just a generally easier

20:39 market to pursue. So I'll start there. Now, with regards to like, how do you approach these folks? Or what

20:45 kind of I think that there's a few different strategies. I think I think that from a if you're

20:51 a first time entrepreneur, or you're just getting started, like being willing to roll up your

20:56 sleeves and do custom development for organizations is a great way to just get your foot in the door.

21:01 A lot of times, there are a lot of individuals in the healthcare system,

21:07 doctors and administrators and all kinds of care providers. They have lots of ideas about

21:13 how they could create a better experience for patients or create a more efficient system for

21:18 delivering healthcare. But they don't have a team of software engineers or they don't have budget.

21:23 And so I think partnering with them and helping build prototypes,

21:27 build proof of concepts, build the ones goes a long,

21:31 long ways to for you as the entrepreneur to like learn about the space,

21:35 learn about the opportunity. But then in terms of that relationship,

21:38 building goodwill and getting access and just like learning.

21:42 When I think about my first two or three years in healthcare,

21:46 I sat in a room with radiologists like daily, just like watching how they use the software.

21:53 And doctors, like they love to complain about software. They'll just like complain,

21:58 which is great because that's like, you're getting the most direct feedback about like,

22:04 what do they want?

22:05 And what makes them angry? What makes them happy? And I can't tell you how many times I've had a

22:09 doctor be like, do you believe this? Watch how many times I have to click one, two, three. And

22:14 they're just like, they're going through and they're just like, they're angry because the

22:19 product was never really optimized for them. And you know, I, if you think about something like Gmail and

22:26 if it took 20 clicks to like compose a new email, yeah, I'd be angry. And like, that's how they feel

22:32 about the tools that they use every day. And so I think just getting access to the providers,

22:38 getting access to the doctors and being part of that conversation. So you can hear their thought

22:43 process. You can hear like what they want, what they like, what they don't like.

22:48 And there's the Henry Ford thing of if you ask customers what they want, they would have asked for a mechanical horse, but they really wanted a car.

22:54 And so I think that you as an entrepreneur and you as the engineer, you have to kind of apply a bit of a lens and a filter to what they're saying and then translating that into something that's achievable or realistic or something that's possible to build.

23:08 But before any of that, like just getting part of that conversation, it's so necessary

23:13 and so important.

23:14 And the vast majority of people in the healthcare system, they have thoughts, they have ideas,

23:19 and they want to talk to tech people about like how they could go about it.

23:23 And so I think just kind of rolling up your sleeves,

23:27 getting through the awkward conversations and talking to those folks is,

23:30 is very, very valuable and you can learn a lot very fast.

23:34 Thank you so much for highlighting this. Wow. And, and, you know,

23:38 circling back on what you said earlier about, you know,

23:40 a lot of folks that try to go generic and not build too close or custom

23:44 relationships because then, Oh, we become a dead shop, whatever. you know, a lot of folks to try to go generic and not build too close or custom relationships

23:45 because then, oh, we become a dead shop, whatever.

23:47 But actually like in that space

23:49 is when you uncover things you didn't know

23:51 and there's opportunity.

23:53 So thanks for highlighting that.

23:54 And for a market like this that you operate in as well,

23:57 is there any prediction you could make,

23:59 you know, for the end of this decade

24:01 about how things will evolve in your space?

24:05 That's a really good question.

24:07 Predictions.

24:08 So I think that there's a few trends that have been consistent now for a while.

24:17 So one, when we were getting started 10, 15 years ago,

24:21 cloud was a dirty word.

24:24 People at that time did not really believe in healthcare

24:28 data in the cloud. And they wanted everything in legacy servers, on-prem, like physical security.

24:35 In healthcare, they often talk about the castle and moat model, where the idea was,

24:40 we have too much software running inside the hospital. We can't secure everything.

24:46 So what we're going to do is we're going to physically secure the building.

24:49 And so everything inside the building is, it's kind of a free for all, but we're going

24:53 to put a moat of like VPNs and air gaps and restricted networks, et cetera.

24:59 And so everything, you know, we're hip compliant inside the castle, but no one's getting in or out.

25:06 And that model broke down pretty badly when there was a lot of consolidation.

25:11 So hospitals are buying other hospitals and now they need to integrate with each other.

25:15 Doctors want to bring their iPhone into the hospital and they want to be able to look at patient data on their iPhone when you can't do that unless everything's on the same network.

25:22 And so that went through a really large transformation

25:25 over the last 10 or so years.

25:27 And I think we're still early in that.

25:29 There's still a lot of software that is very legacy servers, on-prem.

25:35 It's in a rack in the closet, in the basement.

25:38 And there's a lot of benefits to moving things towards cloud.

25:40 And if anything, the story has flipped so hard

25:43 that now the story is we don't

25:45 trust people's local devices. We only want cloud because if it's on their laptop and their laptop

25:51 gets stolen, now that's a breach and we have to report it and dah, dah, dah. Versus if everything

25:56 is very cloud centric and a device has gone fine, we'll just like, you know, kill the accounts and

26:01 that device is utterly useless. So, so I think it feels weird

26:07 to say cloud is a trend because like for outside of healthcare, cloud is kind of like, that's a

26:11 foregone conclusion, but I think it is still, there's still a lot of opportunity there.

26:17 Interoperability, the US government is, has made a lot of very smart moves about pushing healthcare interoperability and healthcare data

26:28 standards in an effort of improving interoperability and improving kind of data

26:36 portability, so to speak. So for example, imagine you went to, this happened to me once,

26:42 I was on a skiing trip and I had a knee injury and

26:45 I got an MRI at the ski facility. And then I went back home and they're like, well, you need to get

26:50 another MRI because we can't get access to your first one. And so, okay. It's just, it feels like

26:58 waste, right? It's, it's, it's waste of my time. It's a waste of my money. It's a waste of resources.

27:03 And that was largely because

27:05 this data portability and data interoperability didn't exist. And the US government recognized

27:09 this, that it's like a massive amount of waste and it's just inefficiency. So this data portability

27:15 is kind of pursuing this further. And we're still early in that. They're not only at the

27:21 kind of technical level, but also in the incentives level that there's now financial incentives for

27:26 being able to share more data to reduce the amount of duplicated tests and

27:30 duplicated scans, et cetera, et cetera.

27:32 So there's still a lot of opportunity just enabling that data standardization,

27:37 data interoperability. I put security in that camp as well.

27:41 The U S government is just getting more and more sophisticated about their

27:43 security requirements.

27:47 And that creates a lot of opportunities to help organizations just get buttoned up

27:50 and compliant on all the security requirements.

27:53 And then there's all chat GPT and AI.

27:59 There's been a lot of talk about

28:00 in healthcare for a long time.

28:03 And I think that doctors have had a pretty

28:06 natural skepticism for machine learning and healthcare. I think that the situation is

28:17 changing. And so first, before we have some kind of super intelligent intelligence, even if you

28:24 just had traditional machine learning techniques, there's a lot of opportunity.

28:27 I think most of the skepticism comes for a diagnostic purpose, right?

28:30 The idea of, oh, you're going to a patient will walk in and a robot will diagnose them and order meds, order tests.

28:37 I don't think that's realistic in the short term, but there's a whole bunch of other stuff in health care that is very automatable,

28:50 There's a whole bunch of other stuff in health care that is very automatable, such as patients sending in messages about questions about billing, questions about insurance.

28:56 I need to reschedule this appointment. Oh, I need to update my medical history. Those things are not diagnostic. They're not clinical. There's no medical decision making.

29:00 But there's a massive amount of human cost that goes into just all of that, like support

29:06 logistics to keep a hospital running, to keep a doctor's office running. And every single one of

29:10 those is an opportunity to apply machine learning or to apply like automation to make things easier.

29:17 I worked with one medical, which is one of the largest primary care providers in the United

29:21 States. And the machine learning team there, like the,

29:25 some of the biggest wins were parsing faxes. There's still a massive amount of just paperwork

29:30 that's going across the wire because there's not interoperability. So just being able to look at a

29:36 fax and identifying which patient this is for, like, this is like a kind of trivial machine

29:40 learning exercise, hugely valuable though, because that's a lot of time and energy.

29:45 And it's an error prone process when humans are doing it manually. So very, very simple thing.

29:50 And I think that we're still early days of applying machine learning at that level.

29:55 And then, OK, now chat GPT and more sophisticated language models are coming around and it creates

30:02 all kinds of new opportunities. And I wouldn't be so brave

30:06 to predict what that will look like, but I think it's an easy prediction to say it will look like

30:12 something. There's opportunity there. When we're spending so much money on healthcare, doctors are

30:18 so burned out with dealing with the administrative workload of healthcare. They just want to practice

30:24 good medicine. They spent years of their life going to med healthcare. They just want to practice good medicine.

30:27 You know, they spent years of their life going to med school.

30:29 They have all this amazing intelligence.

30:31 They don't want to waste a bunch of time dealing with scheduling hurdles.

30:33 Like they want to do good medicine.

30:34 And I think that there's an opportunity

30:37 for a lot more automation in that space.

30:40 Amazing.

30:41 Thank you so much for all these insights.

30:43 And so the megatrends that people can expect, you know, you said is moving from no cloud on premise to cloud,

30:49 interoperability, data standards, portability, and then AI.

30:54 Quick note, because I know it can be a massive topic to discuss,

30:57 but even before going to the machine learning applications over this portable available data,

31:03 you know, even steps before that.

31:05 For medical researchers, what can they expect or hope for in terms of how available data

31:11 will be and, you know, how things will change in terms of publications?

31:17 Great question.

31:18 And some really interesting movements there.

31:22 The historical way was, well, the historical way was very, very tiny data sets.

31:29 And so there's lots of papers that get published or research projects that are conducted where

31:35 they're looking at population sizes of like 50 to 100 patients. It's like tragically small

31:41 because it was so difficult to get access to data or to, to do meaningful partnerships where you can get more, more data.

31:48 So a couple of trends there,

31:50 one is on better anonymization frameworks so that you can, you know,

31:55 work with groups, get like filtered data.

31:59 Anonymization in healthcare is a very tricky topic because to,

32:04 to very safely and robustly

32:06 anonymize data is a hard problem.

32:08 And this is potentially another application for machine learning, but just the act of

32:13 anonymizing it.

32:14 And it does not take too many data elements to uniquely identify an individual.

32:20 And that's what makes the data anonymization process so hard. Especially when

32:25 you include something like a job title and a zip code. Like you can very quickly window that down

32:31 to a single individual, but just those two data elements and you have like, you know, all of the

32:35 other elements, the obvious ones of name, date of birth, like those are gone, but just zip code and

32:40 job title. Sometimes there's only one director of healthcare services in a certain zip code.

32:45 And like, that's an individual. So that's a hard problem, but there's tooling that's getting

32:49 better. So maybe the kind of next evolution from anonymization is the synthetic data in healthcare.

32:55 So this is the process of looking at a real production live data set and using statistical

33:03 methods and other kind of algorithmic processing to generate a derivative data set that has the exact

33:09 same shape and characteristics of the original data set in terms of healthcare

33:12 outcomes and, you know, age and basic patient demographics,

33:17 like gender and age, but also like height, weight, and healthcare outcomes,

33:22 what tests have been performed, all,

33:24 all the different healthcare

33:25 metrics that they can generate that synthetic data set. There's some early players in that space and

33:30 they're doing well. I think there's still a lot more opportunity to kind of move up in sophistication

33:36 and the statistical accuracy of those synthetic data sets. And that will be massive for these

33:42 clinical researchers. If you can have some kind of statistical guarantee that the data is representative, but you can get access.

33:48 And now you have not a population of 50 patients, but a population of 50,000 patients.

33:52 It's just a much more meaningful way to do various initial, like early research that can graduate into like actual patients.

34:02 That is very, very interesting.

34:04 Very interesting.

34:05 Thank you for sharing this.

34:06 And is there today,

34:07 I know we're approaching towards the end,

34:09 just one quick last question.

34:10 Is there someone in this space

34:11 that you so far have,

34:13 you know, maybe looked out for advice

34:16 or, you know, if it's not,

34:17 would you be open to like other founders

34:19 contacting you to, you know, learn some things?

34:22 Yeah, I think that

34:23 as part of the Silicon Valley ethos,

34:26 I think that in general,

34:27 we always want to help those who are coming.

34:31 We've received a ton of help over our time in Silicon Valley.

34:36 Like the startup ecosystem here is amazing.

34:38 And I think that's one of the real strengths

34:39 of Silicon Valley that people are always willing

34:41 to reach out and talk and to help.

34:44 I think for people who are explicitly interested in healthcare, there are a number of organizations

34:49 out there that are specifically focused on helping launch new healthcare startups,

34:56 Rock Health, Startup Health. There's an emerging trend of venture studios. So not just venture

35:03 capitals, venture capital, but also venture studios.

35:07 So they provide not only funding, but also resources, legal resources, compliance, sometimes

35:13 marketing, sometimes software engineering.

35:15 There's a group in New York called Redesign Health that we like to work with.

35:19 They're a big player in this space.

35:20 It's really fun because you have now a kind of a management tier

35:26 of healthcare executives who have deep experience in the space, lots of connections who can help

35:31 point you in the right direction, make introductions. I think that there's an emerging

35:35 trend of these venture studios in healthcare, which I think it's a great trend. It's a great

35:40 opportunity for startups to take away some of the risk that comes with that first one or two years of feeling like you're in outer space and you don't know which direction you're supposed to go.

35:50 That's good. Thank you for those references. And I think this can be helpful to a lot of people.

35:55 And, you know, before I go over the time you made available for this, any last remarks you might like to add?

36:02 And then, you know, could close uh with the med bloom

36:05 description and call to action for people yeah yeah um no i i i think that i i initially was

36:14 very skeptical of going into healthcare as a as a career path and to be honest i've had a uh

36:20 a bit of a love

36:21 hate relationship with it over time

36:26 because it can feel very slow and frustrating

36:29 and sometimes very stifling.

36:30 But over time, I've grown quite comfortable with it.

36:33 And I now really appreciate it

36:36 for the opportunity that exists.

36:38 I think that it's worth being patient in healthcare.

36:41 It can feel slow.

36:44 It can feel over-regulated from time to time.

36:47 But if you're willing to be patient and persist and make the investment in it,

36:51 there's so much interesting opportunity. And there's so much demand for people who

36:56 understand the space for what it is and who have made the investment in who are the key players?

37:05 What are the key challenges?

37:06 How do I navigate the system?

37:07 How do I get things done in this very complex system?

37:11 It's been a great place for my career.

37:14 And so I'm very happy and I encourage others to do it as well.

37:17 With regards to Medplum, if you are launching a new digital healthcare product,

37:22 or you have an existing digital healthcare product,

37:24 and you want to start thinking more about security or compliance or interoperability,

37:28 or how to just accelerate your overall healthcare software development, I think I recommend you

37:33 take a look at MedPlum. We have a lot of great features, great capabilities, lots of documentation,

37:40 example repos, how to get started, how to move fast. If you're not playing in healthcare, but you're a technology enthusiast, please just go over to our GitHub repo and give us a star.

37:53 Like anything that helps us get more visibility, helps improve our reach and gets more access and more viability to these larger organizations. And we just want to keep making progress

38:06 and keep doing whatever we can do

38:08 to promote more innovation in healthcare.

38:12 That was incredible.

38:13 That was perfect.

38:14 Cody, thank you so much.

38:15 Really appreciate the opportunity

38:16 and I hope we can do this again in the future.

38:19 Of course.

38:19 Thank you.

38:20 Thank you.

38:20 All right.

38:21 Cheers.