Subscribe to The Podcast by KevinMD. Catch up on old episodes!

Join Brett Mollard, a radiologist who delves into the exciting future of radiology and the role of artificial intelligence (AI). As we explore the impact of AI on radiology, Brett answers the recurring questions about the potential effects of AI and whether it will replace radiologists. We discuss the profound ways AI is transforming image acquisition, post-processing, workflow optimization, image interpretation, report creation, and result communication in radiology. Brett shares his expert insights on how AI can enhance accuracy, efficiency, and patient care while emphasizing the continued importance of radiologists in the field.

Brett Mollard is a radiologist.

He discusses his KevinMD article, “How AI is enhancing patient care and improving radiologists’ lives.”

The Podcast by KevinMD is brought to you by the Nuance Dragon Ambient eXperience.

With a growing physician shortage, increasing burnout, and declining patient satisfaction, a dramatic change is needed to make health care more efficient and effective and bring back the joy of practicing medicine. AI-driven ambient clinical intelligence promises to help by revolutionizing patient and provider experiences with clinical documentation that writes itself.

The Nuance Dragon Ambient eXperience, or DAX for short, is a voice-enabled, ambient clinical intelligence solution that automatically captures patient encounters securely and accurately at the point of care. Physicians who use DAX have reported a 50 percent decrease in documentation time and a 70 percent reduction in feelings of burnout, and 83 percent of patients say their physician is more personable and conversational.

Rediscover the joy of medicine with clinical documentation that writes itself, all within the EHR.

VISIT SPONSOR → https://nuance.com/daxinaction

SUBSCRIBE TO THE PODCAST → https://kevinmd.com/podcast

RECOMMENDED BY KEVINMD → https://kevinmd.com/recommended

GET CME FOR THIS EPISODE → https://earnc.me/DEaq7P

Powered by CMEfy.

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast. Get CME for this episode by clicking on the CME link in the show notes. Today we welcome Brett Mollard. He’s a radiologist, and his KevinMD article is titled “How AI is enhancing patient care and improving radiologists’ lives.” Brett, welcome to the show.

Brett Mollard: Thank you, Kevin. I appreciate you having me on.

Kevin Pho: We’ll get into the article in a little bit. First off, briefly share your story and journey to where you are today.

Brett Mollard: Sure. So I’m a diagnostic radiologist. I did my residency at the University of Michigan in Ann Arbor, and I was chief resident there. I ended up doing a dual diagnostic radiology and nuclear medicine residency program; with the switching of our boards, we were able to do that. I subsequently went to San Francisco and did an abdominal imaging fellowship at the University of California, San Francisco, after which I went into private practice, and I’ve been in private practice for seven years now.

Kevin Pho: All right. So how’s radiology been? Tell us about some of the rewards and challenges facing the field today.

Brett Mollard: Sure. It’s been a very interesting journey, I would say. Some of the biggest rewards are when we find new cancers, or when we diagnose things that could be subtle. For example, someone is involved in a motor vehicle collision, and we find a subtle traumatic injury that requires going to the OR. Those are the kinds of things that are the big wins, and the most fulfilling. It’s also nice seeing patients who respond well to chemotherapy and things like that when they have cancer; it’s nice to see those positive responses. But really, I’d say one of the most fulfilling things is when we are able to diagnose a curable cancer. For example, if I can see a subtle pancreatic mass that puts the person into the 5 percent who survive versus the 95 percent who don’t, those are the big wins.

As for challenges, across the medical and health care industries as a whole, essentially, we have growing backlogs. We have a lot of retiring physicians, as we all know. We have an aging patient population, which also means we have an aging physician population, and that results in a lot of imaging, with fewer radiologists to read the studies. So burnout is always an issue. It’s been growing since pre-COVID, it got worse with COVID, and it is still getting a little bit worse right now, with a lot of people having exited the health care industry. So those are the main challenges right now, which also leads to my article down the road.

Kevin Pho: We have medical students who listen to this podcast. If they have an interest in radiology as a medical student, what would you tell them?

Brett Mollard: I have a couple of articles on my blog about that, so that’s a great question. The most important thing is just to get exposure and make sure it’s something you’re interested in. In medical school, radiology is one of those specialties that you don’t really get access to. You’ll see a little bit of it when you’re doing internal medicine and general surgery, but you don’t really get your feet that wet. So I would say do a radiology rotation, or even just wander down to radiology, pick their brains, ask them, and see what their life is like. Then, if you want to go into radiology, just show engagement. Do some research projects and take some rotations, and once you’re doing some rotations, they’ll give you some opportunities to get more involved.

Kevin Pho: All right, let’s talk about your KevinMD article. As you know, AI has been at the forefront, especially in medicine and health care, so I’m super interested in hearing your perspective on how it’s going to affect radiology. Your KevinMD article is titled “How AI is enhancing patient care and improving radiologists’ lives.” So tell us, how did your article come together?

Brett Mollard: It started off organically. As you mentioned, AI is in the news and in conversations all the time nowadays, and people are always talking about how it’s going to change things. I think health care is a great fit for AI, just because it’s pattern recognition and it’s following algorithms. Radiology is similar with image interpretation. There are deep learning platforms out there that can analyze thousands of studies that have diagnoses. I know Stanford, for example, has some programs they’ve worked on that can read chest X-rays, and they can diagnose appendicitis. We have all of the imaging out there, so we have the pictures, and then we have surgical proof and we have pathology. So there’s all the information out there for algorithms to learn how to read studies. They’re not necessarily going to be perfect, but I think they are going to make a big impact on essentially what we do.

There are also a lot of other ways that AI can essentially make us more efficient, more accurate, and just better radiologists overall. The efficiency is going to be important because, as I alluded to before, we have these bad backlogs, and they’re going to continue to build as the patient population ages and our physician population retires out. So we’re going to need to be able to work more efficiently to provide excellent care.

Kevin Pho: So what’s the current state of AI? We have these generative AI models, from Google Bard to ChatGPT. Give us some practical examples or stories of how AI is helping your daily life right now.

Brett Mollard: Sure. Essentially, I like to look at it as an exponential curve. Right now we’re still on the low part of the exponential curve, and it’s going to explode over the next decade or so. But AI has already entered radiology; over the past probably five or 10 years, it has been trickling in.

One example is workflow and worklist automation. As a radiologist, trying to figure out what study to read next is wasteful and time-consuming, and I might make the wrong decision because I don’t see the next ER case. So that’s one area where it already exists. The technology is already there where it can prioritize your workflow. You put in all the different key parameters, so I know my ER turnaround time is 60 minutes, or the inpatient turnaround is one hour, two hours, four hours, and it can organize our workflow to ensure that we’re reading the next case appropriately. That way, some cases aren’t going to linger for too long, and we can make sure we get to them in a reasonable fashion.

There are some other ones that are quite nice as a radiologist. This is not something that most people will benefit from, but one example: I read a lot of cancer staging and restaging. For the restaging, we’re always comparing a nodule today to a nodule from three months ago or six months ago. Instead of having to scroll on both series to find that nodule, there’s some software that automatically links the different studies, so I just scroll one of them, the other one automatically scrolls, and that nodule or that liver mass is going to be within probably one to three slices. That makes it a little bit more efficient. It makes it so radiologists will probably be more inclined to do accurate measurements, because it gives them that little bit of extra time where they’ll feel a little bit less crunch. So those are some of the big things.

Kevin Pho: You mentioned that in radiology, of course, pattern recognition and algorithms play a role in terms of your reading. So how good is AI in terms of doing preliminary reads? Are we there yet in terms of AI doing some preliminary reads?

Brett Mollard: Yeah, that’s in very early stages right now. There are tons of companies, like startups, that are creating these. There are some that are offering pneumothorax detection and various chest X-ray detections, so there is some commercial stuff out there right now. As for how good it is, I haven’t tested that out yet, and we can discuss barriers down the road, but that’s going to be one of the barriers: for clinicians to feel comfortable using AI.

One of the other issues, too, is that you don’t want to become too dependent on AI. For example, if it says there’s no pneumothorax, maybe I’m not going to look as closely for a pneumothorax on that study, whereas I’ll have to remind myself that I have to look at the apices and make sure I’m looking for a pneumothorax closely on every single study. But there is some stuff out there. It’s still very early on; again, we’re at the bottom of that exponential curve, but it’s going to explode.

I know for the Stanford stuff that I mentioned a little bit earlier, they’ve been working on this for a while. They’ve been doing this for probably at least five or 10 years, and I think once more stuff comes out of Stanford or other academic centers, there’ll be more trust from the community to adopt it. One of the barriers, like I was alluding to, is that to adopt something that’s going to affect patient care, we have to be pretty comfortable that it’s not going to negatively impact patient care.

Kevin Pho: I do internal medicine, primary care, and I use these generative AI tools to help me write my notes. Now, is there a similar role in radiology in terms of generating your reports, in terms of using AI to assist you?

Brett Mollard: Thank you for bringing that up. In my article, I do mention this a little bit, too. There’s something called Rad AI that we use. Essentially, what it does is it reads our radiology report after we dictate it, and then, when we create our impression, it reads our report and creates an impression based on it. It’s not perfect, but it does save a ton of time. At least for me personally, it makes me more efficient. It puts stuff in the impression that I may have forgotten, because maybe I got interrupted three or four times during that study, or it was a really complex study with 10 or 15 different things going on. So it helps with reminding me of that, and it sometimes can help with diagnosis. I’ve been surprised by some of the impressive calls that it’s made, but there are also times when it’s a little bit less impressive.

Again, that’s still early on, but I think we’re going to see a lot of benefit from that. We have the benefit that there are millions of radiology reports it can review. It also can learn from you, too, so it can read 1,000 of my reports and see how I like to dictate things, and I’ve noticed the impression has evolved over time to sound a little bit more like how I dictate.

Kevin Pho: So let’s talk about steps going forward, in terms of trends that you see. Right now, you mentioned that we are early on in that exponential curve. Do you see a point where AI can replace radiologists? Do you see your job eventually coming under threat because of the improvements AI will have?

Brett Mollard: That’s one of my favorite questions. I don’t think anytime soon. Never say never, but I feel like we have a couple of decades before it gets to that point. There are a couple of pieces to that. One, they are super early. I know this stuff is going to go up exponentially, but it’s so early on in the process. Another thing is that all of these different AI tools are piecemeal. One company has this little niche, another company has that little niche, and so for it to get to that point, essentially somebody is going to either buy up all of those companies, or they’re going to have to work together to be able to do something that replaces us. So I think we’re several decades off before that would happen, if ever.

Another thing that we have going for us, and this is a little sarcastic, is that I don’t think those companies want the liability. I don’t think they’ll want to be sued if their AI software misses stuff, because it is going to miss stuff. You can talk to an amazing radiologist who’s 30 or 40 years into their career, and they still see new stuff every day. A lot of things can present many different ways, where you’ll see a new presentation of a common thing that you’ve never seen before in your 30-year career. So between all of those things, I think we’ll probably be safe. It’s probably more profitable for the companies, too, just to have a continuous subscription service rather than having to be in charge and take ownership of the imaging interpretation.

For anyone who is overly concerned about that, there’s always interventional radiology and mammo. Breast imagers do biopsies, and IR docs do things that robots would be required to do, and we’re a long way from that. So when in doubt, you can always do interventional radiology. They still frequently read a lot of diagnostic studies. So that’s always the safest thing: Just do interventional radiology. But I think we’re going to be safe.

Kevin Pho: We’re talking to Brett Mollard. He’s a radiologist. His KevinMD article is titled “How AI is enhancing patient care and improving radiologists’ lives.” So, Brett, for radiology residents, or even current radiologists, how can they prepare themselves today for what is sure to be a little bit more of an AI-infiltrated world going forward?

Brett Mollard: That’s a great question. I’ve actually been thinking about this a little bit. When you’re talking about Bard, ChatGPT, those things, it’s going to change. If someone is in a writing class and they’re using those to write, they’re not really necessarily learning how to write; they’re going to be too reliant on the AI. So I think that’s a similar analogy: We still want people like residents to dictate their own impressions. We’re still going to want residents to go through that process themselves, just so they’re not going to be overly reliant on AI. But we do want to give them the exposure, so that when they go out into the real world, they’re aware of it, they’re familiar with it, and they’ll be able to incorporate it into their practice. So I think we don’t want them to use it as a crutch, but we want them to be familiar with it so that they can integrate it into their reading.

Kevin Pho: And my final question: Tell us some of your take-home messages that you want to leave with the KevinMD audience.

Brett Mollard: Sure. I think the biggest take-home message that I have is that we need to be open to AI. I know there’s going to be a lot of resistance to allowing it to enter health care, but I think, for the most part, it’s just going to benefit us, and we are going to have that crunch. Burnout is already at a very high level. I know you’ve had some people talk about burnout on the podcast, and with the aging population and the aging physicians, it’s just going to be more and more stress on the health care field. So really, the only thing I think that we can do is hope that AI technologies and solutions are going to come out that make us able to do our jobs more efficiently. If there’s AI that will allow you to do paperwork faster, for example, or Epic notes faster, like you were talking about with your dictations, that’s going to allow us to do more of what we actually trained to do, which is medicine and patient care, and hopefully that stuff will become a little bit more passive.

So the biggest thing is to be open to it. I kind of like “trust but verify.” So don’t just go in headfirst. Make sure everything is vetted and tested out. Always do a trial before purchasing something, so that you can see if you like it and if it meets your standards. But the main thing I would say is just be open to it and give it a try, because I think we’re going to be reliant on it in the future.

Kevin Pho: Brett, thank you so much for sharing your time and insight, and thanks again for being on the show.

Brett Mollard: Thank you so much.