Radiology teams are facing rising imaging volumes, persistent staffing shortages, and growing burnout, making workflow-focused AI more important than ever.
In this episode, Dr. Jeff Chang, Co-Founder and President of RAD AI, discusses building generative AI solutions that help radiologists work faster while maintaining their existing workflows. Jeff shares how RAD AI started by automating parts of radiology reporting, then expanded into patient follow-up and AI-powered reporting tools that reduce dictation time and support more consistent care. He also discusses why successful healthcare AI must fit naturally into clinical workflows, the lessons RAD AI learned while navigating early startup challenges during COVID-19, and how generative coding could accelerate innovation across healthcare.
Tune in to learn how workflow-driven AI can reduce radiologist burden, improve follow-up care, and support the future of medical imaging!
Resources:
[00:00:03] [SPEAKER_01] Hello, everyone, and welcome back to the podcast. Saul Marquez here. Today, I have the privilege of hosting Dr. Jeff Chang. He is with RAD. They're doing some fascinating work in the healthcare space, and today he's here to talk about it. So, Jeff, such a pleasure to have you here. Welcome to the podcast.
[00:00:24] [SPEAKER_00] Jeff Chang Thank you. Great to be here.
[00:00:26] [SPEAKER_01] Of course. Let's fill folks in. To begin with, what got you into healthcare? What keeps you here?
[00:00:31] [SPEAKER_00] I had a kind of a weird story where I started really young, and so I actually had no idea what I was doing early on. So, I went to college at 13, med school at 16. Jeff Chang Oh, you did? Jeff Chang Yeah. Jeff Chang Oh, so you were one of those young kids in college. Oh, man. So, I had no idea. Like, I was in as a chemistry major and preferred not to do a PhD. And then once I was in med school, I was deciding between ER, trauma, and radiology, and just really liked the technology of radiology, how quickly it was advancing. And so, I decided on radiology.
[00:00:57] [SPEAKER_01] Jeff Chang Very cool. Thanks for sharing that. Yeah. I don't think anybody knows what they're doing. Jeff Chang Even in college. Jeff Chang Sure. Jeff Chang I don't think you were alone, despite your age. But very cool. I always admire really bright folks like yourself. Jeff Chang Thank you. Jeff Chang Thank you. Jeff Chang The way that you can move markets is awesome. And so, now you're with Rad AI. Talk to us about the company. What do you guys do? What's the focus and how you're different?
[00:01:20] [SPEAKER_00] Jeff Chang Sure. Yeah. So, we started Rad AI in 2018. So, it's been about eight and a half years. We were lucky in the timing of the company was right when Transform Models first came out. So, the attention paper came out during the 2017.
[00:01:58] [SPEAKER_00] Jeff Chang Jeff Chang
[00:02:26] [SPEAKER_00] Jeff Chang So, the impression section is conclusions, the summarization, and follow-up recommendations customized to each individual radiologist's language. Jeff Chang So, they would get to that section. It would automatically fill out in their language. It would save them that time. Jeff Chang So, they would dictate about a third less, reduce the number of words they would say every day, and then save about an hour per nine-hour shift. Jeff Chang So, that product over time became really popular. It's in use by about two-thirds of all US health systems, nine of the 10 largest radiology groups. Jeff Chang And so, since then, we've rolled out more products. So, our second product was a patient follow-up solution to make sure patients come back for their follow-up exams.
[00:02:53] [SPEAKER_00] Jeff Chang So, say, for example, you have potential cancer or aneurysm that's getting larger. You need to get it followed up every year or every three months or six months. Jeff Chang And so, we make sure that follow-up happens. So, improving quality of patient care. Jeff Chang And then our third product that we released about two years ago is the entire radiology reporting solution using AI to help you generate the entire report more quickly. Jeff Chang You can say a couple words, a couple lines. It'll automatically create the reporting, your format, your templates, your language.
[00:03:16] [SPEAKER_01] That's awesome. So, you've got these three different sections. I love that you guys started with workflows. Jeff Chang Then you've got to close in the loop with patients. And then Voice AI makes progress and you're like, hey, we could find a way to really get to the story faster and help you do it in a note that's customized to how you like to put it out.
[00:03:36] [SPEAKER_01] Jeff Chang Awesome product suite. You were at it for 10 years. So, you were in the thick of it. I was going to ask you for these night shifts. Was it coffee or Red Bull? Or what was your drink? What was your drink of choice?
[00:03:48] [SPEAKER_00] Jeff Chang I mean, I drink a lot of coffee, but at night, I actually have no problem staying up. I have trouble switching back. So, yeah, I can stay up all night. And it got really tough. In the last two and a half years, I doubled the number of shifts as a previous startup. And so, I was working extra for that startup to help fund the startup. Jeff Chang And so, I would do like three weeks of nights in a row. When we're doing 21 nights in a row, your brain just turns into mush. Don't really miss those days.
[00:04:10] [SPEAKER_01] Jeff Chang Yeah, man, that's tough. And I love that you spent the time doing the work. There's nothing like somebody that's been there and experienced it. That also has that tech background or works alongside folks with the tech background to come up with the solutions.
[00:04:27] [SPEAKER_01] Jeff Chang I feel like today more than ever, workflow is where it's at. A lot of this software is at risk. It's headless, you know, these agents, but you guys started there. So, what inspired that beginning? And did anybody fight that start?
[00:04:43] [SPEAKER_00] Jeff Chang We were lucky. It was interesting new technology. We were deciding between a couple different products, some involving vision, all to save time. But we decided on one that really just utilized generative AI. And at the time, we were building all of our own models. Jeff Chang We built our own T2T. We built our own BioBert, so on and so forth. Early in the innovation of Transformer Models, we built everything from scratch. And then over time, we moved to building on top of open source and so on and so forth. So, I kind of have seen the evolution of generative AI over the years. Of course, now moving much faster.
[00:05:11] [SPEAKER_00] Jeff Chang Now it's very interesting that you gain a lot from understanding the workflow really, really deeply. And so since we work with tens of thousands of radiologists all across the country, and we've seen it for almost a decade in terms of all the different workflows that people use, we're really good at building products in terms of this is exactly how each of these different workflows need. Jeff Chang And with authentic coding, you end up being able to prototype really quickly and making things work end to end really quickly.
[00:05:33] [SPEAKER_01] Jeff Chang Yeah, and that's really key. And that's spot on. Those workflows are the differentiator today. Because you put an agent on a broken workflow or a workflow that is not what it should be, you're going to get a bunch of frustrated physicians and you don't want that. Jeff Chang That's a quick way out the door. So that's great, man. And so as you have been in the market and making these advances, what are you finding is top of mind for folks right now?
[00:06:02] [SPEAKER_00] Jeff Chang I think in radiology, people are just burnt out, right? The volume rises by four to 8% every year. And the number of radiologists can't go up because we train right now 1,050 residents per year. And back when I trained 20 years ago, it was 950 residents, right? So that just does not rise. Jeff Chang So as a result, every one of us is reading more and more studies every day. And so you need technological assistance to help you become faster while also not feeling like it's burning you out as much.
[00:06:25] [SPEAKER_00] Jeff Chang And so they really need things that just work with their existing workflows, right? Especially as you start moving into more critical workflow, it has to fit exactly what they need, right? Jeff Chang So it's meet them where they're at, because they're used to using solutions that have been around for 15-20 years, has to have all of these existing features. Jeff Chang And then you can start providing new features that help them save more time and so on and so forth, new AI features they can try out if they want to. Jeff Chang But still has to, the base has to be exactly what they're used to, right? Jeff Chang And that's what we always really focused on, right? Jeff Chang Our first solution, we created a solution that was zero click, right?
[00:06:55] [SPEAKER_00] Jeff Chang It had required no action from the radiologist that automatically generated it at exactly the right time based on where they were. Jeff Chang And it's customized to their exact language. Jeff Chang And so it was really easy to get used to, right? Jeff Chang Radiant Radiologist got used to it within like a minute or two, because it just imitated their workflow exactly. Jeff Chang That's awesome.
[00:07:11] [SPEAKER_01] Jeff Chang That's great. Jeff Chang Jeff Chang
[00:07:34] [SPEAKER_00] Jeff Chang
[00:08:03] [SPEAKER_01] Jeff Chang
[00:08:34] [SPEAKER_00] Jeff Chang Jeff Chang Wow. Jeff Chang That was a big turning point for us. Jeff Chang And then because that was still really early on before we'd even really kind of commercialized, right? Jeff Chang We had a lot of pilot customers but no actual revenue at the time. Jeff Chang And so it's scaled up so much since then.
[00:08:48] [SPEAKER_01] Jeff Chang Man. Jeff Chang You know what? Jeff Chang You guys are probably thinking, I've been there or holy smokes, how did they make it? Jeff Chang And these stories are all around us. Jeff Chang And it's great to take inspiration from Jeff and his team and what they went through. Jeff Chang Because when you have belief and you surround yourself with people that believe in that core mission and what you're doing, the people will come up. Jeff Chang And if you're clear about what you need, you'll get it. Jeff, you've probably read like the algorithm.
[00:09:18] [SPEAKER_01] Jeff Chang I forget the guy's name, but it's Elon Musk's right-hand man at Tesla as they were ramping it up. Jeff Chang He's basically like, man, we were on like three months of cash at all time. Jeff Chang So it's not for the faint of heart. Jeff Chang And that's why I really love that entrepreneurs that make it and are successful, they deserve every bit of it, especially when it's game-changing technologies. Jeff Chang So that's awesome, man. Jeff Chang Thank you so much for sharing that.
[00:09:42] [SPEAKER_00] Jeff Chang It's an inspiring story. Jeff Chang There's always so many existential moments, more early on when you're a solo young startup, but happens all the way through.
[00:09:50] [SPEAKER_01] Jeff Chang Totally. Jeff Chang Hey, so if you had to name one trend that's going to change healthcare as we know it, what do you think that is?
[00:09:57] [SPEAKER_00] Jeff Chang I think the thing that's been top of mind has been agentic coding, like how quickly things are evolving. Jeff Chang I've been using so much SolveXtrahi and before that 5.4Xtrahi. Jeff Chang And of course, the team uses a lot of quad code and so on. Jeff Chang And its ability to allow non-engineers to prototype things very quickly. Jeff Chang
[00:10:27] [SPEAKER_00] Jeff Chang
[00:10:56] [SPEAKER_00] Jeff Chang
[00:11:10] [SPEAKER_01] Jeff Chang Jeff Chang Jeff Chang Jeff Chang
[00:11:28] [SPEAKER_00] Jeff Chang Yeah, it's, I really don't know.
[00:11:31] [SPEAKER_01] Jeff Chang That's a lot of censor. Jeff Chang If you had to though, if you had to, like where do you think this is going?
[00:11:38] [SPEAKER_00] Jeff Chang So in the next year or two, we're getting towards prelim report generation from images, right? Jeff Chang We've been working with partners on that as well. Jeff Chang And it's just that the FDA hasn't figured out how they're going to clear it. Jeff Chang So generating the entire prelim report, first for chest x-rays and MSK films, and then for various other for CTMR ultrasound. Jeff Chang That will gradually happen over the course of the next three, five, seven years. Jeff Chang What happens beyond that in radiology is really hard to say. Jeff Chang There are structural, so the radiologist will always play a key role because not just because of technology, Jeff Chang But because there's both reimbursement and malpractice concerns, right?
[00:12:08] [SPEAKER_00] Jeff Chang So CMS will only reimburse if the radiologist does the final read. Jeff Chang Malpractice company or insurance covers that radiologists, not companies. Jeff Chang Jeff Chang So I don't really know what it looks like. Jeff Chang We're mainly trying to make the experience better in the meantime, but we don't know what will happen in five years or will happen in 10 years.
[00:12:34] [SPEAKER_00] Jeff Chang There's a lot of, I'm very worried about AI disruption across many different industries, right? Jeff Chang Across the global economy. Jeff Chang There's a lot of countries, economies that have already been disrupted severely by, for example, the replacement of call centers and so on and so forth. Jeff Chang There's so many different. Jeff Chang
[00:12:49] [SPEAKER_00] Jeff Chang
[00:13:30] [SPEAKER_01] Jeff Chang
[00:14:00] [SPEAKER_00] Jeff Chang
[00:14:01] [SPEAKER_01] Jeff Chang
[00:14:30] [SPEAKER_00] Jeff Chang So I'm also on LinkedIn.
[00:14:34] [SPEAKER_01] Jeff Chang Outstanding. Jeff Chang Outstanding. Jeff Chang There you have it, folks. Jeff Chang With Rad AI. Jeff Chang They're just moving and shaking in this space. Jeff Chang If you've heard of them, it's just an opportunity for you to dig into the newest stuff that they're up to. Jeff Chang But if you haven't, your chance to dig in. Jeff Chang And if something you heard on today's episode resonates maybe with your friend or a colleague, Jeff Chang Hit forward on this podcast. Jeff Chang Make sure they get it because that's how we make the word spread and improve outcomes. Appreciate y'all tuning in. Jeff, thanks for being with us.