Disruption in Dentistry: Using Artificial Intelligence in Dental Care

EC:Let's just start with you giving us a high-level overview of what AI is and what led you to research and develop an AI product specifically that supports dentistry.
FH:Awesome. Yeah, let's dive into it. So AI, or artificial intelligence, is kind of everywhere, right? And overall it's a not well understood term. And it's really more a category of specific moderates which can do specific tasks. And you can imagine AI being an overarching category. And then under AI, you have machine learning, maybe that's also a term you've heard. And machine learning is really pattern recognition. So you have inputs that can be the shows on Netflix you're watching. That can be any imagery. And then you have a trained system, which is the algorithm, which essentially is like equations. And then you have a defined output. And depending on the input, you have an output. And to give you a really good example, for instance, the machine learning example would be Netflix. When you have a new account on Netflix, you have to type in the top 10 TV shows you really like, that's the input.

And then they have a machine learning algorithm, which correlates to all the other users on Netflix who like similar TV shows and what other TV shows they're watching. So now you have, as the output, recommendations of five other TV shows you might like, right? So that's a simple process of AI and everything is input to an output and machine learning would be for instance, Netflix, and then a subcategory of machine learning would be deep learning, which you maybe have heard as well.

Deep learning now uses even more and more data. And it becomes a little bit harder to understand how the input to the output correlates. But one example would be then, for instance, Alexa or Siri, on your phone where deep learning, because it has massive amounts of data, because voice, et cetera is a lot of information. And then it can tell you as the input like "Siri, how's the weather like?" And then it tells you the temperature. And then a subcategory of deep learning then is computer vision. And computer vision is, has takes now as an input images. Dissects every single pixel of this image. And as an output, for instance, defines decay. So for us, as an example would be dental x-rays, which we take as an input. We have a machine learning system or computer vision system, which has learned on tens of millions of x-rays what, for instance, the pattern is over decay of a cavity of other diseases and as an output at max dose. So that's the basic concept of AI and you can literally apply it to any industry.
EC:So you actually kind of in a way answered the next question that I was going to ask. So, I'd love to actually hear a little bit more, but the next question was going to be with all this data manipulation, if you will, in all these areas, what role is AI playing in dentistry and maybe more specific, how can a dental practice use AI to both increase the quality of patient care and the operations as well, the financial side?
FH:AI can and will play a very significant role in dentistry. Actually, my prediction is that while AI has already penetrated some other healthcare areas like radiology, et cetera, AI will probably have even a deeper penetration in the dental markets in five years from now. Why? So dentists in of itself, they are a clinician in a sense that they are a diagnostician, they are a surgeon, they are a primary care physician and they're a business person all in one. Right? And you saw the rise of DSOs because they don't have enough time to satisfy all of these four roles at the same time. The business component in terms of sourcing material and so on has been tremendously successful from DSOs to take this off your burden. And another piece where AI plays now into the game of dentistry is supporting the dentist and being a better diagnostician and a clinician.

What I mean specifically by this is that AI will support the dentist in the diagnostic and doing x-rays and identifying diseases in the resulting treatment plan, taking in lots of data and identifying based by all these circumstances, what would be the appropriate treatment plan. Then even in the operations, you are discussing this in terms of submitting claims to insurance companies or doing better workflows because right now lots of steps are manual in terms of charting patients getting the documentation together, lots of things which could be automated, right? So across the value chain, if you rethink from a new patient over the diagnostic treatment planning and documentation, and then submission is where everyone AI in a data-driven approach can support this. Now very specifically you ask about the quality of care, right? So, our research at VideaHealth here has shown over and over again, is that over time, the AI can be better in narrow defined tasks than an average dentist.

And that means that it can identify more accurately disease such as decay, such as endodontic issues, such as periodontal issues, than the dentist. So it misses less, but it also has less what we call false positives, meaning that we diagnose something as something and we do a treatment, but actually we shouldn't have done a treatment. So it can really help on both fronts. So that will increase the treatment care tremendously.

And then in terms of the financial operations for a dentist, first of all, it is advisable and good for them to identify more appropriate disease and do appropriate treatments, right? It will also increase the trust between patients and dentists, because you essentially have an ad hoc opinion. You see in dentistry, a lot that patients go to other dentists to get a second opinion, but then they don't return to the first dentist to get the treatment. They're just stuck with the second dentist, right? So here you have now an ad hoc second opinion while the patient's in a chair. And then there are also other just efficiencies in terms of charting documentation claim submission, where you can reduce the workload a lot for a dental practice.
EC:And that's a huge piece right there, again, going back to what you were talking about the reason why they were so interested in DSOs is taking off a lot of that administrative type burden so that they could focus on the clinical. This would make them so much more efficient in the clinical. Efficient and accurate, but yet still being able to deliver that personal patient care and keep that trust and keep the patient in their chair, not even allowing them to go to somebody else's chair for, like you said, that second opinion. You touched on this a little bit also, I think, but what about larger impacts? Do you think that there are larger impacts that you could see from a public health perspective from the use of AI?
FH:Yeah. I mean, I think we will see a huge benefits of it. And that's one of the main reasons why I jumped into this game, because I really think that in dentistry, it is aligned to on one side major successful AI business, but on the other side to actually elevate patient care as well as public health. So what strikes me over and over again is that in dentistry, almost all disease, a hundred percent is preventable. If you would identify it early enough or if you would have preventative measures, right? That is not the case in many other health domains. In dentistry, it is. We've been showing in studies with our partners, led the PTSOs, research institutions or universities, that AI diagnosis much more accurately, as we discussed earlier, but also much earlier. We for instance identified decay six years earlier than some of the dentists.

So there were patients coming into the office and after six years, they are like, "Okay, great. Now we need to do a two surface filling," and they've never identified it and our AI picked it up six years beforehand. So I think what we'll see is that we will catch disease much, much earlier. Right? I think another important aspect is can we use technology to drive down overall costs to give more access to oral health care to patients. Because 30%, depending on the statistics, they don't even go to a dentist. So we need to expand that. And then we will see a reduction in long-term treatments, such as implants, et cetera. So that's on one side and I think that's very exciting. On the other side, we will also see more and more of the research being done on the link between oral health and systemic health.

So I originally come from Germany and you hear that from the accent, I guess. And there's no differentiation between dental insurance and health insurance, et cetera. What we will see is that there might be ways of linking dental data and healthcare data to then flag suspicious systemic health conditions, such as hypertension, diabetes, all diseases, which are closely leading to the oral health of periodontal disease, et cetera. So I think we will see even there an expansion then of oral health going into the overall healthcare.
EC:So then basically using AI to, yes, in this moment, you may be focusing simply on the mouth, on dental, but yet the data points that you're collecting are really also looking and taking your whole body into consideration and looking for everything, any links potentially to anything else, health related.
FH:That's right. That's right. And I think one of the biggest button next year is kind of like the overarching topic in healthcare interoperability, right? So how can you link these data sources to one another? It's actually not so much a technical challenge. Sure, this is a heavy lift to have algorithms understanding this, but the main point is can you link the dental practice management data with the health care data, right? And there are lots of research collaborations going on. Epic is trying to move into the field of dental as well. And I say we will see the big change in the next maybe five years. That's enough.
EC:Okay. That's a lot of good information. If I am actually thinking about utilizing this sort of technology in my practice, what are some good starting points and how would they go about kind of growing from there?
FH:We have a lot of DSOs, also private practices, even insurance companies, et cetera, who reach out to us all the time about using the technology. And the best way is just to either email us at info@videa.ai, or go on our website, videa.ai to contact us. And then a client success manager will get in touch and essentially what this would mean is you could install our software and try it out for two weeks and it would identify disease. And most likely, you like it a lot because you see all of the sudden that you catch things you maybe have caught otherwise not. And it's important for us to elevate patient care. And that's a good starting point, right? Like that becomes then, to be honest, honestly, a standard of care. In every practice we've been working with, it became the new standard of care.

And after that, there are lots of things how you can onboard on further insights. And just to foreshadow, this is imagine you have an AI in your office, which is linked to your practice management data into your imaging data. And now not just while the patient in the chair, it helps you to diagnose better, but it can also retrospectively go through all your data and identify patients where maybe disease has been missed. You can call these patients back and say, "Hey, look, we need to do actually a cleaning. We need to do a scan route playing procedures," et cetera. It can even help you retrospectively to reactivate patients.
EC:Wow. And something that you said kind of right at the beginning of that. You made a reference to DSOs as well as the smaller individual practices. This is, I mean, the way you look at this and the way that you're marketing this is this isn't just for the big DSOs that are out there. This can be for the single doctor practice in any town, USA with, 2,000 patients and so forth. This can be for them. They can implement this, right?
FH:Yeah, that's right. That's correct. I think it's for everyone. I think DSOs are particularly interested in it because they have very young graduates from dental schools with little experience, so to say, and the AI introduces a standard of care. That said, I think one of our highest adopters are just private practices who really want to do the best for their patients. Right? And who are also looking for new technical differentiators, maybe across other practices, et cetera, but really who wants to do the best at their patients and also work with the patients on it. It's a very obvious technology once in the practice, because it's used for patient communication.
EC:Are you seeing any dental schools utilizing this as a teaching tool and something that's going to be coming out that they will actually have their hands on in practice?
FH:Yeah, absolutely. We actually partnered up with ATSU in Arizona. They've been using our AI to train the dental students for the past year. It helps a lot for them to learn how to diagnose, how to treatment plan, and it's kind of a gamified version of okay, you can compete against the AI. And what research has shown us actually. And that's also over and over, and that's kind of the beauty of this. It's not a Terminator story. It's more like a hybrid story that the AI has a certain, we call it accuracy. The human has a certain accuracy, but together they're actually the best.
EC:Yeah. That's fantastic. That’s great because now these students are going to come out with this knowledge of artificial intelligence, how to use it, and they're going to be looking for it. They're going to be looking for it in practice. If the practice that they go to work for doesn't have it, they may encourage the use of it. And if they come out and start their own practice, they're going to be wanting it because it's what they've been trained on. And like you said, now you've got the human side and you've got the automated side and you put those two together and it's a pretty amazing diagnostic tool.
FH:Absolutely. That's why we are predicting that in five years, it will be the standard of care and oral health. And I think that's also fascinating, right? Imagine you're going to your local dentist and from now onwards, there will always be an AI being a second opinion right there, right then, especially in industries, such as dentistry, where you unfortunately, sometimes have a lot of friction between patients and provider.
EC:I agree.
FH:I think this will be a tremendously helpful tool. And that's why there's also so much more value in this than, for instance, for radiology AI products. They are absolutely important as well to identify skin cancer and mammography and breast cancer. But here it's really in the face of the patient. It's really in the forefront. It's not in the back of a dark radiology room.
EC:Right. The patient actually gets to see it in action and see what's happening right there.
FH:That's right.
EC:Any final thoughts around this? Any final thoughts that you want to impart to our listeners about AI and its use in dentistry?
FH:I think that's that everyone who is interested should reach out. And I think there are some obvious first tracks, we just talked about, patient care, et cetera. And then the opportunities beyond that in terms of clinical insights, patient communication, et cetera, are tremendous. I'm really hopeful and positive that this becomes a new standard of care. And also very excited because it's one of the few fields where really AI and data can do something very good. So very excited about this and encourage your listeners as well to be on the lookout for it.
EC:No, I'm excited too. This is amazing. I mean, I really am. Well Florian, I want to just thank you again very, very much for your time today. Thank you for sharing your insights, your knowledge and what I can feel as a passion for the technology, but also a passion for what the technology can do for us, for people, not just for a business, but for the actual care of people. So thank you very much for giving us your time today. Thank you. And thank you everyone for listening to the EisnerAmper podcast, visit EisnerAmper.com for more information on this and a host of other topics and join us for our next EisnerAmper Dental Dialogues podcast.

^{Transcribed by Rev.com}