Latest / Becker’s Healthcare Podcast / Alex Dresner on AI, Radiology and the Future of Technology-Enabled Care

Transcript

Alex Dresner on AI, Radiology and the Future of Technology-Enabled Care · Becker’s Healthcare Podcast

0:02
- The strategic agenda facing hospital and health system
0:04
- leaders right now is not a short list.
0:07
- Growth in a consolidating market, sustainable margins in
0:09
- a reimbursement environment that rarely moves in your
0:12
- favor,
0:12
- a tech investment cycle that demands both urgency
0:15
- and discipline,
0:16
- and a workforce story that is still being
0:19
- written.
0:20
- Eker's fourteenth annual CEO CFO roundtable brings together
0:23
- more than 1,500 executive level attendees and nearly
0:27
- 400 speakers across four days of sessions that
0:30
- are designed to help c suite leaders benchmark,
0:32
- pressure test, and sharpen their priorities.
0:35
- Not in the abstract, but against the real
0:37
- decisions they'll face heading into the next year.
0:39
- It's the kind of event you leave with
0:40
- a clear sense of where you stand and
0:42
- what comes next.
0:44
- So join us November second through the fifth
0:46
- at the Hyatt Regency in Chicago,
0:48
- and come ready to do the hard thinking.
0:50
- Register on our events page at beckershospitalreview.com
0:54
- by clicking on the events tab in the
0:56
- upper right.
0:57
- This is Scott Becker with the Beckers HealthCare
1:01
- podcast.
1:02
- I'm thrilled today to be joined by a
1:03
- brilliant entrepreneur
1:05
- and thinker
1:06
- who's making a difference,
1:08
- more so than I've seen in so many
1:10
- people that I talked to in in the
1:12
- radiology,
1:14
- and and and screening area and just fascinating
1:17
- what they are doing and what he is
1:19
- doing,
1:20
- with Alex Dresner,
1:22
- who's the CEO and one of the founders
1:24
- of Imagine.
1:25
- And and, Alex, can you take a moment
1:28
- and and tell us a little bit about
1:30
- yourself
1:31
- and about what you folks do? Yeah. Absolutely.
1:33
- So I'm Alex Dresner. I'm one of the
1:35
- cofounders and CEO of Imagen Technologies.
1:38
- We help health care organizations
1:40
- deliver faster and more accurate radiology diagnoses to
1:43
- every patient, and we do this by combining
1:46
- subspecialized
1:47
- radiologists
1:48
- with industry leading,
1:50
- imaging technology and AI that we build in
1:53
- house.
1:54
- Now to understand why I'm here, you have
1:56
- to understand the personal stakes.
1:58
- I've lived the devastating reality of diagnostic error.
2:01
- I actually lost my father to lung cancer
2:03
- because it was missed on an x-ray, and
2:05
- I also personally survived metastatic testicular cancer
2:09
- after an initial error on an ultrasound.
2:12
- Now those experiences
2:13
- weren't just medical events. They were, for me
2:16
- especially, moments of profound vulnerability.
2:19
- And when I was correctly diagnosed with thyroid
2:22
- cancer just a couple years ago,
2:24
- the difference in experience, the
2:26
- the fear, the clarity, the the speed to
2:29
- diagnosis and treatment was night and day, and
2:32
- that is the experience that every patient deserves
2:35
- in my opinion.
2:36
- Now today, seventy percent of patients require medical
2:39
- imaging for a diagnosis.
2:42
- That what that does, it makes radiology
2:44
- a utility of sorts for the hospital, similar
2:47
- to water or electricity.
2:49
- But currently, that utility is broken.
2:52
- There's a massive worsening shortage of radiologists in
2:55
- The US, and it's expected to hit forty
2:57
- percent over the next seven years.
2:59
- And the result is that patients wait weeks
3:02
- for results, and mistakes, unfortunately,
3:04
- are,
3:06
- inevitable.
3:07
- And it's also why 92%
3:09
- of health systems are currently exploring a change
3:11
- in their radiology provider.
3:13
- Now turning now to Imogen,
3:15
- our vision here is to transform radiology
3:18
- from this artisanal,
3:20
- high variability,
3:22
- and high error specialty
3:24
- into a semi automated service that is repeatable,
3:27
- it's reliable, and accurate every time,
3:30
- akin to how the industry standardized the, the
3:33
- clinical lab tests.
3:35
- Now
3:36
- we're achieving this vision by combining our own
3:39
- radiologists with our own proprietary
3:41
- AI to deliver an end to end clinical
3:44
- service to to hospitals and health systems across
3:46
- The US.
3:47
- The model first starts with great radiologists.
3:50
- These are experts deeply committed to patient care.
3:54
- But to find them, we've we've adopted almost
3:56
- like this, moneyball style approach where we rely
3:59
- less on academic
4:01
- pedigree in the resume
4:03
- and more on data driven correlates of success
4:06
- that we've identified through years of testing these
4:08
- radiologists.
4:09
- Now let's take the the VA, for example.
4:12
- We found that the VA produces some of
4:14
- the most accurate radiologists in the country.
4:17
- It's not an accident. It's because these radiologists
4:20
- at the VA are looking at images of
4:22
- critically ill patients all day long.
4:24
- Those reps, if you will, the constant exposure
4:27
- to to both common and rare pathologies,
4:29
- builds a diagnostic instinct that is hard to
4:31
- replicate in other care settings.
4:34
- And we look for that kind of experience,
4:36
- not just, the resume.
4:39
- And so then it's also about pairing those
4:41
- great experts with great technology.
4:43
- Our proprietary AI at Imogen is really focused
4:46
- on two areas.
4:47
- At the point of care, our AI our
4:50
- FDA cleared AI acts as a second set
4:52
- of eyes highlighting suspected conditions as the radiologist
4:56
- works.
4:57
- Now importantly, this has led to a seventy
4:59
- three percent reduction in misses of life threatening
5:02
- conditions,
5:02
- things like lung cancer that took my father.
5:05
- We also use AI in a second way,
5:07
- and that's behind the scenes.
5:09
- Our AI reviews every report to catch the,
5:12
- if you will, the silly errors like a
5:14
- mislabeled body part,
5:15
- ensuring that clinical accuracy before the report ever
5:18
- reaches the referring physician,
5:20
- and the patient. Now the market of hospitals
5:22
- and health systems and other health care organizations
5:24
- has responded
5:25
- incredibly well because this model works.
5:28
- We serve at Imogen two point five million
5:30
- patients today across 28 states, and we've had
5:33
- ninety eight percent lifetime customer retention
5:35
- while still continuing to grow top line at
5:37
- more than a 100% year over year.
5:40
- We are ten years into this journey, but
5:42
- we're really just, if you will, at the
5:43
- starting line. The future of radiology, in our
5:46
- opinion, just isn't about sort of more AI.
5:49
- It's about the fundamental shift towards a
5:52
- reliable AI enabled operational model.
5:55
- And what Imogen hopes to become here over
5:56
- the next five years
5:58
- is really be recognized as the sort of
6:01
- definitive benchmark
6:02
- for quality, reliability, and care in US radiology.
6:06
- So, Alex, remarkable
6:09
- what you've got
6:10
- started and going and how it's going.
6:13
- You mentioned one of the stats you mentioned,
6:15
- and there's so many different things I wanna
6:16
- dig into here.
6:18
- One of them is this concept that ninety
6:20
- two percent of health systems are trying to
6:22
- figure out what to do about the radiology
6:24
- providers and so forth.
6:26
- Can you talk about that a little bit
6:28
- further?
6:29
- Yeah. So,
6:31
- this is based on a survey,
6:33
- that we executed recently of a 100 of
6:36
- the largest health systems across the country.
6:39
- And it is a pattern across The US
6:42
- right now that hospitals and health systems are
6:44
- struggling to hire radiologists, struggling to keep up
6:48
- with the demand
6:49
- for patient imaging, and, ultimately,
6:51
- those errors, those delays in patient care are
6:54
- really having a big impact on downstream, not
6:57
- just the diagnosis, but the treatment and management
7:00
- of those patients. And it's become untenable for
7:03
- a lot of these hospitals
7:04
- and health systems,
7:06
- such that they're, you know, almost on mass
7:08
- now looking to to swap out their legacy
7:10
- radiology provider,
7:12
- which is in most cases typically
7:14
- another outsourced radiology group, often private equity backed,
7:18
- that is struggling to attract and retain its
7:21
- own radiologist to keep up with care.
7:24
- Thank you. And
7:26
- when you see
7:27
- all the
7:29
- implementation
7:30
- of AI and screening and so forth,
7:34
- Now radiology
7:35
- was the original use case for telehealth
7:37
- back in the day
7:39
- when there wasn't enough radiologists to cover remote
7:42
- or other location, just not enough for the
7:44
- amount of places that need to be covered.
7:46
- Over the last several years, it's been you
7:48
- know, we're again back to huge shortages of
7:50
- radiologists,
7:51
- both
7:52
- interventional and diagnostic.
7:55
- What does AI mean in the long run?
7:57
- You know, at some point, people thought AI
7:59
- would decimate the radiology profession,
8:02
- But at least what I see currently
8:05
- is there are so many images being done,
8:08
- and and they still need a human in
8:10
- the loop on literally everything it seems like
8:12
- or at least anything that's complex. And
8:15
- that's probably never gonna fear word choice.
8:17
- But what is the long term impact on
8:19
- the labor versus technology or labor and technology
8:24
- for
8:25
- diagnostics
8:26
- and interventional?
8:27
- I mean, I still think there's gonna be
8:29
- an absolutely critical role for radiologists in the
8:31
- future. And in many respects, I think, like
8:33
- you see in other industries,
8:35
- that role will actually become stronger through time.
8:37
- I think the role of AI is
8:39
- it's effectively taking the bottom third or bottom
8:42
- half of what a radiologist does off their
8:44
- plate.
8:46
- Today,
8:47
- a radiologist really does a few things. They
8:49
- are looking for visual patterns in an image.
8:53
- They have to synthesize those patterns and figure
8:55
- out what they actually mean, and then they
8:57
- have to write the report that is communicated
9:00
- downstream to the referring physician.
9:02
- The radiologist and the clinical expertise that they
9:05
- spend years and years developing is best suited
9:08
- to answering the question of so what after
9:10
- what they've seen in in an image. AI
9:13
- can do the highlighting things in a medical
9:14
- image. It can even do the
9:16
- the report writing,
9:18
- and what that does is it allows the
9:19
- radiologist to spend more time focused
9:22
- on intervening and providing care to more patients
9:25
- and and really answering that question of, okay.
9:27
- Here's the findings, but what do they actually
9:29
- mean, and what are the appropriate next steps
9:31
- for the patient? So,
9:33
- like with other applications in health care,
9:35
- as well as in under other industries, I
9:37
- think AI for radiologists is just gonna, help
9:40
- them, continue to operate at the, you know,
9:43
- highest possible part of their license.
9:46
- Thank you. And when you look at sort
9:48
- of the next
9:49
- few years,
9:51
- you've had this tremendous growth.
9:53
- I understand that you've got a tremendous number
9:55
- of radiologists that are working with you. What
9:58
- are you most focused on and excited about
10:00
- over the next year to two years?
10:02
- Yeah. So
10:04
- our focus here is really around making Imagen
10:06
- the best place for a radiologist to to
10:08
- practice their craft. And and we often talk
10:10
- about AI as if it's sort of the
10:12
- protagonist in the health care story, but, for
10:15
- us, it's really the radiologist is the hero.
10:17
- And and our job here is to give
10:20
- these radiologists the the tools to perform at
10:22
- the top of their license.
10:24
- And so, you know, if you look over
10:25
- the past eighteen months or so, I'm I'm
10:27
- really most proud of how we've moved from
10:29
- sort of this theoretical to to almost operational
10:32
- excellence. It's one thing to build a cool
10:34
- piece of technology. It's it's really another to
10:36
- deploy it, so seamlessly that it actually reduces
10:39
- the cognitive load rather than sort of adding
10:42
- another,
10:43
- screen or widget for the radiologist to look
10:45
- at and process.
10:47
- And and you can sort of see those
10:49
- results in our culture. We're seeing,
10:51
- ninety two percent annual radiologist retention, which is
10:55
- significantly outperforming the industry average of, 78, 79%.
10:59
- We're growing organically by 20 radiologists a month,
11:02
- making us one of, if not the fastest
11:04
- growing organic radiology group in The US.
11:07
- And that retention and growth really tells me
11:08
- we aren't just selling a service here. We're
11:10
- building a home for the best talent. And
11:12
- and, arguably, we're just getting started here. So
11:15
- over the next twenty four months, my focus
11:17
- is on taking
11:18
- that, you know, quote, unquote, hero experience to
11:21
- the next level. We're building technology that through
11:24
- time and with the right regulatory approvals, we'll
11:26
- shift the radiologist from proverbial,
11:29
- writer to to editor by leveraging AI that
11:32
- can,
11:33
- not just detect visually detect the findings on
11:35
- the image, but pre generate reports,
11:38
- and do so across,
11:40
- nearly every condition in an X-ray, CT, MRI,
11:44
- and ultrasound,
11:45
- which for our clinical practice and is sort
11:48
- of typical of most clinical practices
11:50
- amounts for about eighty five, eighty six percent
11:52
- of all imaging that we do.
11:55
- And then from there, our goal is really
11:56
- to build
11:58
- an undeniable evidence base that proves that this
12:01
- semiautomated
12:02
- AI enabled practice isn't just sort of this
12:05
- aspiration, but it it really is the and
12:07
- should be the new standard for how we
12:09
- can and must deliver care across The US
12:11
- going forward.
12:12
- Thank you. And and
12:14
- give us a sense of
12:16
- you've had this incredible experience of
12:19
- founding this company, I think, with cofounders.
12:22
- But what advice,
12:25
- one, to entrepreneurs
12:26
- founding a company, any advice there?
12:29
- And then two, the health system leaders
12:31
- or big payers trying to figure out the
12:33
- radiology service and how they deal with radiology.
12:36
- Any advice there for those two audience? One,
12:39
- entrepreneurs,
12:40
- two, health systems payers. How do they think
12:42
- about, you know, what they how they should
12:44
- look at radiology and these kinds of clinical
12:46
- services?
12:47
- Yeah. Absolutely. So so maybe a few to
12:50
- start with a few pieces of advice for
12:51
- founders.
12:54
- So I'd say that first,
12:56
- persistence is really the absolute baseline.
12:58
- If you aren't willing to endure the no's,
13:00
- you really shouldn't be in the arena. But
13:03
- persistence alone isn't enough. It has to be
13:05
- coupled with this sort of relentless commitment to
13:08
- the customer's mission. Now let me give you
13:11
- sort of this example from 2020. We were
13:13
- a a young company,
13:15
- as young companies do. We were about four
13:17
- months away from running out of runway,
13:19
- and we had pinned our future on winning
13:22
- a commercial contract with a single massive national
13:25
- health system.
13:26
- The CEO had been unavailable or at least
13:28
- unwilling to meet with us for a year.
13:30
- And when we we finally got a fifteen
13:32
- minute window to meet with him in person,
13:34
- I was facing this sort of kinda crazy
13:36
- logistical impossibility. He was 1,500 miles away from
13:39
- me. I had twenty four hours to meet
13:41
- with him, and I couldn't fly because we
13:43
- were in the middle of COVID.
13:44
- There was no vaccine yet, and and my
13:46
- wife was, in the middle of a high
13:48
- risk pregnancy.
13:49
- Now we didn't make excuses. I I chartered
13:51
- a tour bus. We I drove twenty hours
13:54
- straight, pulled into their corporate offices, and
13:57
- despite walking out of what was actually Maroon
13:59
- five sixty foot, you know, fully blacked out
14:02
- tour bus, I should have chartered for less
14:04
- than the cost of a plane ticket because
14:06
- nobody was was touring during COVID.
14:09
- But it wasn't really about the bus. It
14:10
- was about signaling to the CEO
14:13
- of this very large national health system that
14:15
- we were the the type of partner who
14:17
- would walk through fire to deliver for them.
14:19
- And when the CEO saw that level of
14:21
- commitment, he didn't just give us the meeting.
14:23
- On the spot, he gave us a a
14:25
- large commercial deal and an equity investment.
14:27
- And that really taught me in the early
14:29
- days, you don't just sell a product, you
14:30
- sell your your commitment to your partner's success
14:33
- as well.
14:34
- The thing second thing I would say to
14:35
- founders is you build resilience,
14:38
- but you have to pair it with,
14:40
- intellectual honesty. Now most founders fail because they
14:44
- fall in love with their initial solution rather
14:46
- than the problem they're trying to solve, and,
14:48
- Imogen almost fell trapped to that, exact same
14:51
- problem. When we started, we were
14:53
- a SaaS company selling software. We had hoodies,
14:56
- jeans,
14:57
- and a great pitch deck. And we'd heard
14:59
- 200 no's before we realized that software that
15:02
- hospitals didn't want another software tool. They wanted
15:05
- a partner to solve their operational crisis.
15:09
- And we had to be willing to kill
15:10
- our initial business model to save the company.
15:12
- And and those were humbling difficult days, but,
15:16
- you know, the the graveyard of AI start
15:18
- ups is full of these teams that were
15:20
- too stubborn to pivot.
15:21
- We survived at Imogen
15:23
- because we were more obsessed with solving the
15:25
- health care problem problem than being right about
15:28
- our initial product.
15:29
- Take take a moment on that because, Alex,
15:31
- I think that's so important.
15:33
- People are so sick of solutions being sold
15:36
- at them
15:38
- versus truly solving a problem.
15:40
- And it's it's almost we work with a
15:42
- lot of AI companies that increasingly we have
15:45
- four deployed engineers
15:47
- so they could really be on-site and close
15:49
- to and responsive to customers.
15:52
- Whereas, if they're just throwing software at them,
15:54
- customers can't integrate it in and don't want
15:56
- to do it.
15:58
- Talk about that sort of complete solution really
16:00
- solving a problem mindset
16:03
- versus a piece. Well, I mean, you you
16:05
- hear it in the words that when you
16:06
- when you go talk to your customers or
16:08
- or prospective customers, I I'll never forget in
16:11
- the the words of one of my favorite
16:13
- CEOs from of of a large health system
16:15
- in the Northeast. He goes, I want a
16:18
- single throat to choke for all my radiology,
16:21
- problems. And, you know, perhaps
16:24
- a a strong message, but I think the
16:25
- essence of that is
16:27
- hospitals and health systems are dealing with any
16:29
- number of challenges on any given day and
16:32
- over the course of what they're trying to
16:34
- achieve over a multiyear period,
16:36
- and they want to be able to have
16:38
- a reliable partner to solve large operational and
16:42
- strategic challenges for them. And I think a
16:45
- lot of founders in particular I see
16:47
- focus
16:48
- from the standpoint of what is the product
16:51
- versus what is the customer, what is their
16:53
- the end need and end solution that's needed
16:55
- here.
16:56
- And I think in general, where that trends
16:59
- for a lot of health care companies is
17:01
- in order to do that, you have to
17:02
- become
17:03
- a service solution and increasingly a technology enabled
17:07
- service solution to be able to
17:09
- scale yourself and deliver a better, faster, cheaper
17:12
- solution through time than relying, you know, as
17:15
- much on technology as you do on the
17:16
- human service component.
17:18
- So it's not just about service, but it's
17:19
- about that technology enabled service model. And I
17:22
- and and just in talking to a lot
17:23
- of my peers, a lot of, health care
17:25
- founders,
17:26
- a lot of folks shy away
17:29
- even recognize that that's how health systems buy,
17:32
- but they shy away from the complexity of
17:34
- health care services, from the
17:36
- medical legal risk associated with it. And it's,
17:40
- honestly, it's a very hard business to build
17:43
- what is effectively two separate businesses until they
17:45
- become one. You have a technology arm. You
17:47
- have a services arm. And through time, you
17:49
- wanna bring those together, but you're effectively building
17:52
- two businesses,
17:54
- to be able to offer something that is,
17:56
- you know, a single,
17:58
- you know, enhanced solution for for your customers.
18:01
- That that that really solves a problem. And
18:03
- I think your your your point is so
18:05
- well taken. I would almost say it a
18:06
- different way. If you're at a company selling
18:09
- into health system,
18:11
- if you really wanna own the problem that
18:13
- the health system has,
18:14
- you need enough lieutenants, enough leaders, enough people
18:18
- that could really own those problems,
18:21
- and
18:22
- that's
18:23
- expensive and hard to build that.
18:26
- The technology
18:27
- is hard to build. It's also hard to
18:29
- build enough leadership team, enough people in an
18:32
- organization that really own problems and solve problems,
18:35
- and you really need both to do this
18:37
- well.
18:38
- And many people don't wanna invest
18:40
- not easy
18:41
- at all to send a lot of engineers
18:43
- after a problem.
18:45
- It's almost a harder problem to integrate that
18:47
- with problem solvers that are working really closely
18:49
- with customers with health systems to make sure
18:51
- you're really fully solving their problems, which I
18:53
- think is what you folks are doing at
18:55
- Imagen, which is remarkable.
18:57
- That that's exactly right. And then that sort
18:59
- of almost dovetails into your point about what
19:01
- advice might we give to health system leaders.
19:03
- Well, first off, I don't, you know, I
19:05
- I think they probably know how to solve
19:07
- their their problems better than I do. But
19:10
- I would focus on
19:12
- make the adoption and integration of AI
19:16
- your
19:17
- partner's problem, your partner on the other side.
19:20
- I think as you're running a health system,
19:21
- the idea of bringing in all this different
19:23
- AI and trying to
19:25
- integrate it and figure out the,
19:28
- best way to make use of it, keeping
19:29
- up,
19:30
- with the speed of evolution of the technology,
19:33
- the the challenges associated with change management,
19:37
- of your clinical administrative staff, and and all
19:39
- the associated issues. Those are
19:41
- big problems and hard problems to solve, especially
19:44
- doing so across so many different specialties and
19:46
- so many different administrative areas. I would love
19:48
- for partners whether it's like a technology enabled
19:51
- service like Imagen or others in other spaces
19:54
- where you're you're partnering with a group who's
19:56
- solving those problems on their side and then
19:59
- just sell providing back to you the,
20:02
- you know, better, faster,
20:04
- cheaper,
20:05
- solution that gives you the benefits of AI
20:08
- without actually all the complexity of having to
20:10
- to integrate it yourself.
20:12
- Thank you, Noah. And I and I and
20:14
- I love that. And it's it really is.
20:16
- The word bespoke solutions is such a wrong
20:18
- word because it really is, but it's almost
20:21
- forward deployed solutions where you're really engaged in
20:23
- solving the problem for people
20:25
- versus sending stuff to them to solve where
20:28
- they've gotta take up so much of the
20:29
- Slack themselves. It it it really is.
20:32
- That's right. It it no. Thank you. Alex,
20:35
- it's amazing what you're doing. I wanna congratulate
20:37
- you on what you what you've come with
20:38
- Imogen and what you're doing. Solving a critical
20:41
- problem. It it it it it seems like
20:44
- as these diagnoses and the tools move faster
20:46
- and faster,
20:47
- we're still gonna need so many radiologists to
20:49
- keep up with that. You know, it it
20:51
- it almost reminds me now of how the
20:53
- law changed years ago when you get your
20:54
- test results, the same time the doctor gets
20:56
- your test results. And if a doctor can't
20:59
- give you guidance on those pretty quickly, it's
21:01
- a very anxious time for a family or
21:03
- person.
21:04
- That's 100%
21:05
- right. Oh,
21:07
- fascinating what you're doing. So impressed with what
21:09
- you're doing at Imogen and what you're growing.
21:11
- Alex, I wanna thank you for taking the
21:13
- time with us today on the Becker's Healthcare
21:14
- podcast. Just a pleasure to visit with you.
21:16
- Thank you so much. Thank you. I appreciate
21:18
- it.