RadNet, Inc. (RDNT) Earnings Call Transcript & Summary

Good day, everyone. Welcome to the RadNet Inc. DeepHealth Artificial Intelligence Call. Today's conference is being recorded. At this time, I'd like to turn the call over to Dr. Howard Berger, President and Chief Executive Officer of RadNet Inc. Please go ahead, sir.

Thank you, operator. Good morning, everyone. This is Dr. Howard Berger calling in. And 2 days ago, we -- and the brand had announced that it had acquired DeepHealth and formed a division for artificial intelligence that will be led by the President of DeepHealth, Dr. Greg Sorensen. Today, we want to talk about RadNet's decision to move more substantively into artificial intelligence as well as the strategy that hopefully, RadNet and DeepHealth will be executing together to accomplish those goals. Ten years ago, RadNet made a critical decision to invest and earn its own information technology platform, which we call, eRAD, which would serve as the backbone for managing a lot of distributed network of imaging centers. The premise of this decision was based on both economic and operating requirements we believe represented an existential necessity. We view this strategy has proven successful as operating RadNet without this capability seems improbable. I believe RadNet and in fact, all of radiology and imaging, is at a similar crossroads. Artificial intelligence will invaluably impact every facet in how the practice of imaging is delivered. Given the size to which RadNet has grown and likely to continue growing, artificial intelligence becomes another existential necessity to maximize operational and clinical opportunities. Over the past year, RadNet and DeepHealth have been collaborating in the use of artificial intelligence for mammography. During this period, we came to know the DeepHealth team, and in particular, Dr. Gregory Sorensen. When RadNet made the decision to make a substantial investment into artificial intelligence, it was obvious that Dr. Sorensen's credentials were uniquely qualified to lead this effort. Clinical research radiologist, business and technology experience, running the North American division in Siemens Healthcare and evolving the detailed effort in mammography AI over the past 4 years. The opportunities which lie ahead in artificial intelligence practically fall into 3 categories: one, clinical accuracy and productivity enhancement; two, business and operating efficiencies; and three, revenue enhancement. Before I turn the call over to Dr. Sorensen, I would like to give some perspective to the category of revenue enhancement. Procedures performed in diagnostic imaging centers generally require a prescription for the particular exam from the referring physician. In addition, for advanced imaging procedures, authorization is also required before the exam can be performed. The only exception to this rule is screening mammography, which the patient can self-refer for annual or bi-annual follow-up. As a result of this capability, RadNet has sought tools to improve compliance and increase volume by direct outreach to the patient population. Since RadNet implemented the Whiterabbit.ai application, mammography volume in most of our regions has increased substantially. The benefit of using mammography artificial intelligence for more accurate and earlier diagnosis of breast cancer is rapidly being validated. Due to recent technology advances, the opportunity to use AI for prostate, lung and colon cancer screening are now possible. Similar to mammography, the potential new screening tools should eventually be adopted by patients and payers as both affordable and necessary, and as a part of population health initiatives, which improve outcomes and reduce costs. Best of use of AI for these screening tools should create a significant benefit, which will help direct patients to centers which utilize this capability. Both RadNet and DeepHealth share this vision and made the combination and collaboration opportunities compelling. I'd like to turn the call over now to Dr. Gregory Sorensen, who will further elaborate on both the circumstances, which led RadNet and DeepHealth to come together as well as his vision of the opportunities for the future in artificial intelligence. Greg?

Thank you, Howard. Hello, everyone. It's a pleasure to be with you on the call today. I'd like to begin by affirming what Dr. Berger has said about the potential impact for artificial intelligence, and in particular, a form of machine learning known as deep learning, to improve the health of women and men. Dr. Berger's categorization of opportunities that is those 3 opportunities, he mentioned clinical enhancement, business operations and revenue enhancement are also spot on. Before diving into the specifics of how DeepHealth's team and technology will address each of these issues, I'd actually like to review a couple of commonly discussed concepts to help set the stage. Let's start with an analogy that's gotten a lot of attention, namely the phrase, "Data is the new oil." In some ways, that's a very apt analogy. But I'd like to say that a better one is that, "Data is the new tight oil or tight natural gas." Getting the value out of the source, requires a lot of work, perhaps not precisely analogous to fracking, but still a very energy-intensive, sophisticated and time-consuming process. If we were to continue that analogy, DeepHealth would be like an oil extraction company and RadNet would be like the Permian Basin, a vast resource with tremendous potential. Bringing these 2 together just enables so much opportunity. This analogy may be even more apt because it's clear that the 2 companies are such a good fit for each other. From DeepHealth's perspective, it's far more efficient for us to be inside the sources of data that we need to do, to the AI that we want to do to seamlessly deal with collecting the data. Something that, of course, is made much easier at RadNet, as you've heard, RadNet owns the IT infrastructure itself. We also have instant access to friendly and cooperative medical experts to ensure that our AI isn't developed in a vacuum. This is something that other imaging companies struggle with from time to time. And from RadNet's perspective, having what you might think of as the best fracking company in the world to show up and offer to help them realize value of all this data is also very exciting. Bluntly put, I believe that the most valuable assets that RadNet has are not currently listed on its balance sheet. The data and clinical scale that RadNet has is so tremendous. The opportunity to unlock that value is just incredibly exciting. So that's one concept that I wanted to discuss. Another concept that is quite common in AI is to say that AI is going to replace radiologists. While I'm a radiologist myself and RadNet works with over 750 radiologist directly, and I can tell you that there's never been a better time to be a radiologist. AI is not going to replace us. AI is empowering us. We at DeepHealth and RadNet expected our machine learning technologies will enable RadNet physicians, meaning our affiliate radiologists, to practice medicine at the very top of their license. We at DeepHealth are building the AI to do the drudgery parts of the job, which will enable our physicians to provide better quality care, while also being more productive. So how are we going to do that? What are some of the specifics? Well, let me open the hood a little bit on our technology and our team to help explore that. As you know, at the core of DeepHealth, from the name you might guess, is deep learning. And deep learning is the name given to a relatively new way to let computers learn to do certain tasks. In our case, the breakthrough is that we can develop deep learning technologies to actually help physicians do a better job of interpreting images. Deep learning consists of taking what is called a model, which is really just a set of equations, sort of linear algebra equations, and adjusting the coefficients and the variables in those equations until the model produces the answers that you're hoping to see. And the great thing is that this adjustment can be done automatically by exposing the set of equations or the model to training examples. So if you want your model to distinguish pictures of dogs from pictures of cats, you could start with the computer model consisting of lots of linear algebra equations, show those equations thousands of pictures that you know are a bunch of dogs and other thousands of pictures that you know are pictures of cats. And you'll end up with a model or a software system does a pretty good job at distinguishing dogs from cats. And in fact, this is no longer so hard to do. And many companies around the world are using systems like this today for everything from software for self-driving cars to patient recognition to designing drugs. But what if you wanted to model to distinguish dogs from wolves? Or in our case, images that were mammograms that show something that might be cancer versus mammograms that shows something that actually is cancer. This is where the hard work comes in. We at DeepHealth have built teams of people, very talented people, to address this question. And they in turn had built tools to start to address this type of question. As with wolves versus dogs, we've carefully identified examples of, on mammograms, breast cancer versus not breast cancer. Not by using the opinions of humans, but by tracking down pathologic evidence for each example, each one of those mammograms. Our team has specifically built software tools, methods and even patented approaches for adapting deep learning to the specific needs of our medical imaging problems. And in the case of breast cancer, because fortunately, only 1 in 200 women who undergo screening mammography actually had breast cancer, the key to getting lots of pictures of wolves is to be sure that you can first collect lots and lots of pictures dogs. That's why we at DeepHealth are so excited to be a part of RadNet because of the large volume of medical images that RadNet generates every year. We have the opportunity to build precise, powerful tools that really do answer important medical clinical questions. So with that backdrop, now let's turn back to Dr. Berger's categories, so where AI can benefit patients and specifically help RadNet do its job better. This first category, clinical accuracy and productivity, is where the DeepHealth team has been focusing its effort in its first 40 years. To illustrate the power of AI and its opportunity for clinical benefit, I'd like to briefly discuss a little bit more screening mammography. Screening mammography clearly saves lives. It's been studied in multiple countries multiple times over the past 3 decades. And since screening has been introduced, breast cancer deaths having indeed declined. But that doesn't mean that screening mammography is a perfect tool. Reading mammograms, I can tell you from personal experience, is a difficult task and it's difficult even for full-time experts. Because breast cancer is fortunately still quite rare, as I said earlier, only 1 out of 200 women who get a screening mammogram, actually have a breast cancer present on that mammogram. But the signs of cancer, those 1 in 200 can be quite subtle. It's actually not a bad analogy, if you go back to the idea of trying to sort dogs from pictures of wolves. Sometimes they can look very, very similar. And because it's so rare, only 1 in 200, that means the other 199 women do not have the breast cancer. All in all, this is very much like the proverbial task of finding a needle in a haystack. And this can be, for us physicians, both simultaneously very tedious, because so many are normal, yet very stressful. Because as an interpreter, you know that there's a cancer somewhere in those 200 women's cases. Because this task can be so difficult, in turn, that means not all doctors are equally good at it. The best radiologist might catch well over 90% of the cancers, but there are some who appear to catch more like 60%. And recent technology improvements in the way we take those pictures has made the task even more difficult. A new technology called digital breast tomosynthesis, which RadNet is a pioneer and a leader in and rolling out, it's sometimes called 3D mammography, does do a better job of showing the cancer on mammogram. But because it's 3D instead of 2D, it converts 50 to 100x as many images for us experts to review, making our job even more difficult. This situation is exactly where AI should be able to help. We at DeepHealth, our team has created a database of thousands and thousands of verified examples of breast cancer to build AI software to identify even subtle signs of cancer. We've been testing this software now with RadNet. And for over a year and -- or I guess, nearly a year, and we've learned that our software can find cancers that human physicians sometimes overlook, sometimes as much as a year earlier. We're now working with RadNet's expert physicians to determine not only how much better at interpreting mammograms the DeepHealth software can allow doctors to be, but also to measure how much more efficient they could be if the software could help them prioritize their work. So that's that first example. Clinical accuracy and productivity, it's very clear that there's some exciting opportunities there. Some really exciting low-hanging fruit. Now the second category that Dr. Berger mentioned of business and operating efficiencies is where RadNet's earlier team, Nulogix, has been working already. And while the breakthroughs in image analysis capture a lot of attention in the press and media, et cetera, many of the same deep learning tools can be applied to nonimaging data and can have substantial impact. One example, the RadNet has in active testing now is a tool that analyzes the dozens, if not hundreds of variables that play in to how to best submit a reimbursement request to an insurance company. This helps RadNet immediately because having a claim rejected by an insurer not only means a delay in cash collection, but it also raises risk of never being paid at all. And of course, another big bonus of these kinds of applications -- these business applications of AI is since they're not involved in direct medical care, they do not require FDA approval. The third category, revenue enhancement, is something that Dr. Berger has already alluded to. If AI can help us lower our costs at RadNet, to do screening exam patients, be those examinations mammograms or lung CT scans, which have been proven to save lives in smokers and former smokers, or say prostate MRI, then we can offer these exams to large populations of patients at an affordable cost and actually improve millions and millions of lives. It's that opportunity, the RadNet opportunity, that makes us so excited to join together with them. In closing, I'd just like to return to the idea that RadNet's best assets are not yet visible on its balance sheet. I've had a chance to work closely with the RadNet team over the past year, and we and my whole team are convinced that not only is the raw material present in the data, but that the entire RadNet management team and clinical teams also understand the possibilities here. And they're eager to work with our highly talented scientists to unlock these assets. Doing this in a business setting is very exciting to us because it sharpens our scientific focus onto the things that are sustainable and valuable to all patients and to all stakeholders. We're very motivated to get going. And with that, I'll close my remarks. Thanks very much for your attention, and I think we're going to turn it now to some questions. Is that right, Dr. Berg?

Yes. Great. Thank you. All right. We will take your questions. And please, operator, give us the first questioner.

[Operator Instructions]

While we are waiting for people to come up with their questions, let me just jump in because I can repeat a few questions that I often get about deep learning. One is -- I'd just like to circle back to this question of will it replace radiologists and are radiologists doomed? There's a famous meme going around about how radiologists might be like Wile E. Coyote that are already over cliff. That question or that statement was made back in 2016 or so. Here we are 5 years later and -- or 4 years later, and the job market for radiologists is fantastic and there's huge demand for our specialty. And that's because so many people want the value that imaging provides. RadNet's, by being focused on imaging, really does bring one of the most valuable technologies to patients. And the opportunity for AI to speed that up and to make it more efficient and to improve its quality, this is what radiologists really are excited about. The human element of actually speaking with the woman to explain what's going on her mammogram, discuss the procedure of a biopsy you might do. These things are never going to go away. Computers are never going to actually consult with patients. At least, I don't see that happening in my lifetime. And so it's really a symbiotic relationship, where the AI and the doctors work together to actually deliver things in a much smoother and more efficient way. Okay. I'll pause here and see if there's any questions coming in.

So it looks like we do have a question from Mitra Ramgopal with Sidoti.

Yes. First, I just want to get a sense in terms of what details that you'll be able to bring to the table, just for example, Nulogix might not have been able to offer. And as you combine the 2 entities, would you say you have the full range of capabilities to be able to drive the workforce efficiencies and lower the cost that you intend to? Or do you need to invest even more in bringing in some additional technologies?

Well, It's a multifaceted question, let me try to break it apart. First off, the team that we had in Nulogix, while very competent, was limited to, practically speaking, 4 people. And their primary function, as Dr. Sorensen had mentioned, was looking at business efficiencies, such as improving our reimbursement operations by reducing denials, getting a better sense of the collections at the front door for co-pays and self-pay, et cetera. And so their functions while -- their capabilities might have been broad, were rather limited. With the acquisition of the DeepHealth team, their focus has been solely on clinical opportunities. And as Dr. Sorensen was describing, primarily in the field of mammography. For the past 4 years, they've been performing that task of bringing a product to market, which we hope to apply for FDA approval later this year, and then be able to start using it clinically and perhaps even then selling it commercially. So the development for the basics of the algorithms that are used to develop these clinical tools, particularly as Dr. Sorensen was saying, we're identifying cats versus dogs. Once you set up those processes, they're more easily adaptable to other opportunities where you might want to use the same basic logic and strategy, and which is what we hope the DeepHealth team will turn to either after or simultaneous with their work in mammography. And as I mentioned in my opening remarks, our particular focus will be for prostate screening and then lung screening and colon screening, all of which, between the cancers that are identified for those in those 4 organs, comprise about 80% of all the cancers diagnosed in the United States annually. So the market opportunity for us to develop these tools is not only very broad and extensive, but it's more important that the understanding of how we want to have these tools with something similar to mammography become adopted by the medical communities, so that patients can get these in an affordable manner, and one where they can self-refer if they have either the clinical background and risk assessment, or it's just as part of screening, for example, like prostate cancer, can be done on a regular basis. So we knew that RadNet or anybody would have to attack this kind of a broad approach, would require substantial resources in assets. Even if we go after the more strategic initiatives that we'd like RadNet to pursue because they need not only to lead clinical and accuracy of our radiologists, they lead to revenue opportunities. These are still only a fraction of the opportunities inside artificial intelligence, and it's quite likely because we have been working with other teams on artificial intelligence that come to us for both our data extraction as well as clinical expertise, that we will be either partnering or collaborating with other teams so that we can further develop in a more expeditious manner the benefits of artificial intelligence inside RadNet and for the community. So I expect that this to be a broad approach. And it's important that we have somebody of Dr. Sorensen's capability, both from a clinical standpoint, a research standpoint and a business standpoint, to be able to lead what I see it as a broad-based initiative on the part of the company. It's also quite possible that we'll be working with some of our manufacturers, the OEMs and equipment that we buy, given the resources that we have in tools both in terms of the amount of equipment that we might have out there and the ability to implement a wide range of delivery of these services as well as having expertise in running the business side of imaging that most of the equipment manufacturers lack. So I've seen the resources, both coming from an expansion of the combination of the DeepHealth and Nulogix exchange that Dr. Sorensen will oversee as well as bringing in other relationships with business entities, whether they be manufacturers or people that are just focused on other parts of the AI opportunities as part of an overall strategy for the company.

Okay. No, great. I was also wondering if maybe you could give us a sense in terms of how does the FDA approval process works for AI. And your sense in terms of getting your product to market, that the kind of time frame we might or should be looking at?

Sure. Great question. I'll take that. This is Greg. The FDA has asserted jurisdiction over medical imaging artificial intelligence products. They consider them medical devices. In past decades, they considered them Class III or devices for certain risk categories. But in the past year or so, they have reclassified most AI tools, like the ones we're developing, to be Class II devices, that will require 510(k) clearance before they can be used in interstate commerce. And so we have -- we at DeepHealth have been spending substantial amount of time in talking with folks at the FDA. We have met repeatedly with them. We've built a good relationship of trust and scientific integrity. And we believe that will be an ongoing part of our strategy, where we will continue to work closely with FDA about the products we want to bring to market. The typical 510(k) process, as you know these days -- well, it does kind of continue to evolve because AI is such a new field. But in general, the FDA requests that you have a, what they call a presubmission meeting with them. You kind of go in and explain what you're thinking about doing. They give you some feedback. That takes a few months. Then you go out and do the work that you've essentially laid out on what you expect to do. That can usually take a few months. And then you submit the material to the FDA and they get a few months to actually review things. So all in all, once you're kind of pretty close to done with your technology and ready to start the regulatory process, that can be about a year, sometimes a little longer. It depends on how good your science is going into the FDA first. To your specific question about RadNet's tools. We will be getting FDA clearance for those AI technologies that the FDA has guided us that require FDA clearance, and we expect that will be a pipeline of things over the years. Just to -- doctor -- your earlier question, just to echo what Dr. Berger said earlier, there are so many opportunities in AI and radiology. And RadNet has such a broad portfolio. It's not like we're a single imaging modality business. That for some things, we're going to build internally at DeepHealth. Some things, it'll just make more sense business-wise for us to buy, if you will, or to go outside and partner with external parties. And so we'll be looking at that closely as we move ahead. It's definitely not a invented-here syndrome here. We're very open to the best technologies where we can find them to help RadNet meet its mission and to help our patients.

Okay. No, that's great. And then just to follow-up a little on that. Obviously, the focus has been on mammography. How easier -- how quickly do you think you will be able to use these technologies on the -- some of the other modalities?

Well, so some of the modalities have tools commercially available already today. And RadNet has been evaluating a number of them commercially, even as we speak. The real question, I think, from a investor point of view and from an owner point of view is, "Well, how valuable are those?" And so that's really the process that I think we'll be doing over the next quarter or 2, is trying to figure out which of these technologies can really make an impact for our shareholders and for our operations versus the kind of more nice-to-have but not necessarily meaningful. And I expect that over the course of 2020, we will identify some things that really will have a meaningful impact. Now in medicine, things take a long time, practices and habits take a lot to change because there's so many people involved. So I wouldn't want to overpromise or say we're going to see things turn on a dime. But the opportunities to identify those, RadNet is really well situated for it because we are -- have such broad operations. We can kind of distribute different opportunities to different teams. And in parallel, identify, I don't know, 5 or 10 different things all -- kind of almost simultaneously. I think we're as well situated perhaps, better situated than most entities out there to address AI opportunities at scale.

Good. Let me to interject something for a moment, Mitra. Greg, can you give some perspective, at least as it relates to mammography AI, where DeepHealth is in the current process of filing for FDA approval?

Yes. So specifically, with our product, we met with the FDA last November, for what's called this presubmission meeting, had a very productive, positive experience there, came out with some very clear guidance from the FDA about how to proceed. We will be meeting with them again next month to get final approval for our execution plan. And basically, the way that works is, you can come up with a -- essentially an agreement with the FDA, a handshake agreement equivalent, where they essentially say, "If you execute the scientific products that you are proposing to us and it turns out with the scientifically proven answers you think it will, then that will support the kind of claims that you're looking for." And in our case, we're looking at DeepHealth and RadNet, not just to have, oh, a claim that says, "We can do a little bit better at maybe interpreting mammograms." But we're very focused on being able to demonstrate to the FDA's satisfaction and our own satisfaction, that the AI actually makes a meaningful difference in women's lives. So we don't yet have clearance to -- of what that will actually look like. I can't give you a specific quote on what the FDA will agree to. But scientifically, I can tell you, we're interested in things like -- something like a physician who interprets a mammogram using DeepHealth software would find cancer earlier than if they interpreted a mammogram without DeepHealth software, that kind of thing. And I know our data suggests that we'll be able to do that. And we think that's a substantially unique differentiator in the marketplace. And that's -- more importantly, it will be something that our referring physicians, see as very valuable for their women -- their female patients. And that's the kind of direction we would go, not just the dollar operational efficiencies, but meaningful improvements in women's and men's lives. And we think that's particularly valuable in the RadNet setting because it's something that we haven't really talked about on this call, but I know came up in yesterday's call and that was, RadNet is unique in that it has capitated lives, or relatively unique. Not many radiology providers can take on capitation. And that opens a whole series of value propositions for artificial intelligence that are going to be possible and even very meaningful. That when you're sort of outside the tent, or not so closely linked, just become much harder to do or monetize. And so there's a bunch of things we're going to be able to do together, be it by working hand in glove and by DeepHealth being part of RadNet and RadNet having its own DeepHealth, if you will. That would just otherwise, just be much harder to do from a business perspective.

And let me add, Mitra, a couple of things before perhaps you have -- if you have another question. First of all, to put better in perspective, even though only 1 out every 200 women whom we do screen mammography and ultimately get diagnosed with breast cancer, in the general population, 1 out of every 8 women, will get some form of breast cancer in their lifetime. So we are talking about a very substantial problem and as Dr. Sorensen suggested that even though there has been great strides in diagnosing breast cancer earlier, it hasn't necessarily reduced the number of breast cancers. It's simply allowing us to diagnose it earlier and with the use of artificial intelligence and other technologies, such as 2D tomography, we should be able to diagnose these earlier and ultimately reduce morbidity and lower the cost of the system. So I think it's important for all of us to understand that this is about scale. And that goes to the heart of RadNet's decision to become more fully invested in artificial intelligence. We are on a trajectory to do about 2 million mammograms a year. And so on an annual basis, that would represent close to 5% of all the mammography performed in the United States, and that's done in, ostensibly, the 5 major markets that we're in. If you wanted to take technology, such as DeepHealth is on the brink of getting the FDA approval from and then purchase that, you would add $1 or $2 per read, per scan, it would cost the company $2 million to $4 million and be an ongoing cost that we would continue to bear regardless of how many mammograms are ultimately done inside the RadNet organization. And we'd only be able to do with that kind of expenditure, really one product, or have one product. Mammography is unique in what we do in that every mammogram that we perform will need to be subjected to artificial intelligence. So this is into medicine, only use artificial intelligence on the patients that you ultimately diagnosed to have breast cancer. It's a tool that will help more accurately and earlier identify those cases. So every mammogram we do will need to be subjected to artificial intelligence review, so both to determine normals from not-normals and then essentially be another pair of eyes to help the radiologist read these more accurately. The same logic could apply and should apply to, for example, prostate cancer. Every man, should they live long enough, is likely to get prostate cancer. The question in front of the medical communities is often, "What do you do for the testing? And when -- what then do you do for treatment?" given that prostate cancer and many of the prostate cancers will not necessarily be something, which ultimately leads to somebody's demise, but becomes a very substantial cost of the system to treat. We firmly believe, and we've done some initial work with this that we can develop a tool that will make prostate MRI screening as easy and as affordable as mammography screening, will substantially reduce the cost of the diagnosis and will allow, I think, the payers to make this every bit as accessible to their membership as mammography is now to women. So I want to put that in perspective because that becomes another tool that we believe will be subjected to every prostate MRI that we do. We just hope to do them earlier, diagnose them faster and more accurately. The other comment I would want to make, and then this is really looking down the [indiscernible] piece is that when we do a mammogram, when we do an MRI, particularly the prostate, there is a lot of information on those scans that are simply discarded because they're either not the reason why the test is ordered. They're just not practically capable of doing. I saw recently an article about a team out of Stanford that won an award for taking MEMRI scans and subjecting them to artificial intelligence and then being able to determine risk assessment for cardiovascular disease in all those patients that had MEMRIs, regardless of whether the reading itself was determined to be normal or abnormal. Essentially, that kind of information is available on every scan or procedure that is done in imaging. And that data, which is another part of the assets, if you will, that any radiology company or provider owns, is something that I think can yield an enormous amount of valuable information. With that, particularly from mammography and prostate, if we were to do that on a screening basis for the population, would essentially be additional information that wouldn't cost anything more to develop then giving the opportunity for artificial intelligence to determine a particular perspective that we're trying to achieve with that diagnostic information. So the perspective of what we're talking about in artificial intelligence is really not just under at one tool, but potentially looking at the spectrum of the impact artificial intelligence can have. Overall, in the delivery of health care, particularly as it is represented in the radiology diagnostic imaging field. I'm sorry, Mitra, if you had another comment or question you'd like to ask, please.

No, that is great. And the final question, and again, this is more down the road. Would the reimbursement be necessarily any difference if AI is used versus not AI?

Well, that's probably, again, going back to my analogy of the $64,000 question, although that's from a number of years ago, that TV series. My guess is, the answer to that, maybe no. I don't know if anybody is going to pay you more because you're doing it. It's more likely to be that you will do more and see more of those patients because you're doing it and providing a value-added to the system that will make you a better provider and a better partner in dealing with the overall health of that patient. Now I could be wrong, but if you look at what's happened over the years in imaging as new technology has been advanced, nobody is running out there and generally paying us more money. The one exception to that is perhaps at this time to be mammography, which does have an additional code that you get reimbursed on above and beyond 2D mammography. But even the number of dollars that we get paid for that are not necessarily reflective of the additional investment that has been made. However, 3D tomography from mammography procedures is now the state-of-the-art. And most people will be expecting you to perform that exam, and if you don't, you're at a competitive disadvantage. I expect the same type of thing to be embraced by the payers and others that will have artificial intelligence as a requirement that if you don't provide it, you actually may get less money. So -- and that even has some historical relevant...

Precedent.

Precedent. Thank you, Greg.

It becomes the new normal.

Yes. And I'll give you the example for that was that when the industry moved to digital radiology, and it was required for reimbursement by CMS, they didn't pay you more if you did it, they paid you less if you didn't do it. So again, perhaps I'm being a little overly dramatic when I use the term existential necessity. But I think we have to look from a historical standpoint, and we have a saying about what's going to happen on the reimbursement side. And we all tend to live in this world, by the way, of still everything is on a fee-for-service basis. But the industry is moving more to population health, alternative reimbursement models, risk-taking, getting paid by better outcomes, if you will, and artificial intelligence will be a critical part of that opportunity moving forward.

Yes. And if I could just echo one thing kind of it's being -- that I want to bring out. The scale of RadNet enables us to have the potential to develop these AI tools. Dr. Berger mentioned that we're going to do, we think this year, 2 million mammograms. That's as many as the entire United Kingdom will do next year. So RadNet can collect data, analyze data, get all those pictures of dogs and wolves or whatever analogy you want to use to answer the most critical and valuable questions at a scale of very few other organizations can even contemplate. We have our own internal AI system -- I'm sorry, IT system. So all the data collection gets streamlined. We have a lot of alignment of interests. Just the opportunity to catapult ahead and really bring these technologies to patients and to make business impact faster. Just really, a really remarkable opportunity that we both saw, and we're just super excited about.

All right. I think we have no further questions at this time.

Great. All right. Well, thank you all for joining us today, and please stay tuned for exciting developments that RadNet hopes to be able to deliver this year and bringing better medicine, which is good business, to the marketplace. Thank you all for your time this morning.

Thank you. Bye.

That does conclude today's conference. We thank you everyone again for their participation.