In this episode of The Golden Age of Orthodontics Podcast, Dr. Leon Klempner and Amy Epstein are joined by Dr. Shankar Rengasamy Venugopalan, Chair of the Department of Orthodontics at Tufts University. Together, they discuss the transformative impact of AI on orthodontics and digital treatment plans. They’ll explore AI's role in treatment planning, emphasizing the need for transparency in AI algorithms and the importance of orthodontists' clinical judgment, as well as some of the limitations of AI. Dr. Venugopalan highlights the biological factors influencing tooth movement and reassures that AI cannot replace human expertise and patient-doctor relationship. Now is the time for orthodontists to embrace, rather than fear, technological advancements to enhance patient care through innovative approaches.

IN THIS EPISODE: 

• [3:30] Meet Dr. Venugopalan's background and expertise in orthodontics and AI
• [5:09] Precision medicine and its relevance in orthodontics
• [10:34] Dr. Venugopalan gives an overview of AI applications in diagnosis and treatment planning in orthodontics. Insights into the current state of AI research and its clinical applications.
• [19:12] The potential for integrating genomics and multi-modal data in orthodontic treatment. Plus, a discussion on the limitations and challenges associated with AI in orthodontics.
• [28:24] Will AI replace the need for orthodontists or serve as a supportive tool in treatment planning? 
• [33:26] It is important to consider the patient’s biology in treatment outcomes and the inherent limitations of AI tools in this regard
• [38:21] Dr. Venugopalan encourages the development of holistic clinical knowledge beyond just biomechanics in orthodontics and expresses hope for the integration of AI in improving patient outcomes.

KEY TAKEAWAYS: 

• AI is currently being applied in various aspects of clinical orthodontic practice, from diagnosis to treatment monitoring. These applications not only streamline clinical workflows but also enhance the quality of care provided to patients. 
• Despite its potential, AI in orthodontics faces several challenges and limitations. One major concern is the quality and diversity of data used to train AI models, as biased or incomplete datasets can lead to inaccurate predictions. Additionally, the integration of AI into clinical practice requires significant investment in technology and training, which may be a barrier for some practices.
• While technology offers numerous benefits, it is essential for orthodontists to maintain their clinical judgment and expertise. Technology should be viewed as a tool that enhances, rather than replaces, the orthodontist's skills and knowledge. As technology continues to advance, the role of orthodontists is evolving to encompass new skills and responsibilities. In the future, orthodontists will need to be proficient in digital tools and data analysis to effectively incorporate technology into their practice. They will also play a crucial role in interpreting AI-generated insights and integrating them into personalized treatment plans. 
• As the field becomes more tech-driven, orthodontists will have the opportunity to enhance patient care through innovative approaches, while also navigating the ethical and practical challenges that come with technological integration.

RESOURCE LINKS

People + Practice  - Website 

Dr. Leon - Email

Amy - Email

People + Practice - Email

Our Golden Age of Orthodontics podcast sponsors make it possible for us to bring you new episodes. Lightforce Orthodontics is revolutionizing orthodontic care with cutting edge custom bracket technology. that offers personalized digital treatment plans that are fast, precise, and uniquely tailored to your patient's needs.

Narrator: The future of orthodontics is evolving and changing every day, but although the way to achieve practice growth has changed, there's never been a better time to be an orthodontist.

Let's get into the minds of industry leaders, forward thinking orthodontists, and technology insiders to learn how they see the future of the orthodontic specialty. How will digital orthodontics, artificial intelligence, clear aligner therapy, remote monitoring, in house printing, and other innovations Change the way you practice.

Join your host, Dr. Leon Klempner and Amy Epstein each month as they bring you insights, tips, and guest interviews focused on helping you capitalize on the opportunities for practice growth. And now welcome to the Golden Age of Orthodontics with the co-founders of people and practice, Dr. Leon Klempner and Amy Epstein.

And I'm joined by my lovely, bright, Intelligent, caring, empathetic daughter, Amy Epstein.

Amy Epstein: I like the intros today. The intros today are superior to the intros normally. Thank you for that. Uh, thanks dad. I'm Amy Epstein. I have an MBA in marketing, 20 years of marketing and public relations experience. And about 10 years ago, we joined.

Speaking of which today we are very excited to well, so my dad mentioned that he's a tufts grad today. So that becomes particularly relevant. He's a jumbo. Um today we're very excited. Hey,

Dr. Leon Klempner: hey, hey, hey, hey.

Amy Epstein: I mean, not only are you the tallest person I know. I

Dr. Leon Klempner: did put on a little weight, but come on.

Amy Epstein: You're, you're also jumbo in other ways.

He completed his dental training at the University of Missouri, Kansas City, earned his orthodontics certificate, and a DMSC in oral biology from Harvard and a PhD in biomedical science from Texas A& M. He is a specialist in genomics and craniofacial disorders, craniofacial development, mineralized tissue biology, and clinical orthodontics, and he's led significant research published widely.

And presented internationally, he's also served as the president of the craniofacial biology group and the International Association of Dental Research and has received numerous awards, no doubt. Um, and today of particular interest, um, in talking to Shankar is the, his understanding of AI as it relates to orthodontics from both.

Dr. Shankar Rengasamy Venugopalan: Thank you. Thank you, Amy, for your kind introduction and thank you, Leon, for having me here. I'm I'm excited, excited to be part of this podcast.

Dr. Leon Klempner: Uh, well, we're excited to have you and, um, you know, uh, we hear about a lot of changes in healthcare and I want to kind of, like, start with kind of a more broader view because I know that.

Orthos in general have mixed feelings about AI and what's going to happen in the future. Some of them are threatened. Some of them are excited. And I do want to dig into some of that, but I first want to talk about just overall trends in healthcare in general. And I know that you've spoken before about, uh, the P4 paradigm in precision medicine.

And just for the benefit of our listeners, could you just explain exactly what that means and what that trend is showing us?

Um, it is becoming more and more Um, relevant these days based on, um, numerous clinical studies. I'll give you an example, right? So you go to a doctor, you have, uh, some type of a disease and the doctor is running a bunch of diagnostic testing and then come to a diagnosis or, uh, a clinical decision standpoint, and then.

So in the context of precision medicine, we think about four P's. The first P is predictive. So you want to have a technology that is highly predictive of a certain condition or a certain disorder. And then the second aspect, so it's not just enough it is predictive, but it also have to be preventive. So, it's, it's predicting at the right point where you can actually prevent the disease.

So the P4 paradigm focuses on these, uh, four aspects. So, uh, think about it this way, how does it relate to orthodontics? 20 years ago, 30 years ago, people tried to develop a sort of a playbook for ortho, right? If it's class 2 division 1 malocclusion, 1, 2, 3, 4, this is how I'm going to treat it. But after practicing a number of years, you come to realize there are a lot more variables that are beyond our control.

So, in the context of orthodontics, the predictive, the preventive, and then the personalized aspect has come a long way, and I think, uh, the P4 paradigm is not just relevant for our general medicine, but it is also very much relevant in the context of, um, orthodontics. Another example would be, a patient A develops severe external apical root resorption, Whereas patient B doesn't.

Why is that? There is a big genetic component to it. Individuals carrying genetic susceptibility for a certain amount of root resorption could range from mild to severe and you're applying the same level of forces to retract the anterior teeth in an extraction case and one responds with severe root resorption and other doesn't.

Amy Epstein: Well, that makes a ton of sense. Thanks for that, that, um, introduction and how it's relevant to orthodontics. So, um, how does AI then come in to support orthodontists who are looking to apply this, uh, P4 paradigm Um, you know, moving forward in the practice.

Thank you. And then we can also focus a little bit on, uh, what is that already in, uh, existence for, uh, clinical utilization. So, um, in the field of orthodontics, especially in the context of applications of AI, uh, a significant body of work has been done in the context of Diagnosis and treatment planning, right?

So, the most basic one is manual cephalometric analysis versus, uh, AI driven cephalometric analysis. Now, most orthodontists, after practicing for five years, they can look at a ceph and, uh, Make a, a skeletal diagnosis. They don't really have to, uh, draw the lines and measure the numbers. You know, the level of accuracy is within one to two degrees.

So you have a certain set of patient comes to your office, let's say skeletal class 3. Now you have to make a determination. Let's say the patient has passed the age of, um, early intervention, and now you have to make a decision is this patient well within the range of, um, camouflage versus the patient really needs, uh, jaw surgery.

So, the benefit of this type of AI application is, so, you know, Leon has practiced for decades. So he may have come to a point where just by looking at the patient based on his experience, come up with a plan that might produce the most favorable outcome. Someone two, three years out of the clinical practice, we try to, you know, as an educator, we try to teach our residents an evidence based approach, but The clinical experience plays a big role, right?

Right. And I'm sure as we, uh, move through this podcast, we'll also talk about what are some of the limitations or what are some of the barriers that prevents translating this research into, uh, uh, a program that could be clinically utilized. And the other area is in prediction, you know, uh, friends of mine from University of, uh, Illinois, they published a paper, they've developed an AI based model system to, to predict cervical vertebrae maturation.

Individuals who are in the middle, right? C3 and C4, you put a group of 10 orthodontists in the room, three will say C3, C4, and then some will say it's between C3 and C4. So, uh, my colleagues in the University of Illinois, uh, College of Dentistry, Dr. Mohamed Elnager and, uh, Dr. Saath El Oredi, they have published a paper, uh, utilizing AI model to make Uh, predictions of cervical vertebrae maturation, where some of these applications, as soon as you take a lateral sef, uh, if the AI algorithm could be integrated into our, um, patient management software, then it could, you know, before you walk to the station, you already have all those details readily available to you.

You take CDs of images and you can compare one time point versus another time point and make interpretations as to how well the treatment is, um, You Progressing and this has a lot of benefits as as as one could realize the patient, you know, the patient may not have to visit every so often as they would otherwise it could limit the number of times the patient actually visiting the provider as long as they are making progress with their.

And this becomes particularly relevant in the context of, um, more and more orthodontists are treating adult patients. Um, and this, uh, AI application also has an ability to type in notes and whatnot. So, If you look at the applications of AI broadly in the context of orthodontics, like I said before, some are in the research and development phase, mostly focused on diagnosis, treatment planning, predicting, and what is right now in the clinical arena is monitoring treatment progress.

It is very well established in medicine. Uh, like for example, If, for cephalometric tracing, you, you, you're mainly, uh, looking at one diagnostic modality of the many that is used by the orthodontist. The future of AI application really requires integrating multi modality of the data, multi dimensional data.

Now, even further down the road, genomics could be integrated. Uh, as part of this, and, um, you know, with, uh, companies like 23andMe and there are, uh, several other enterprises where you can actually go and get your genome sequenced. So all that data is already out there. And, uh, a whole bunch of genes have already been established and associated with, um, skeletal class 2 malocclusion, class 3 malocclusion, enamel hyperplasia.

Then without that information, you may, you may not choose to do a treatment plan that requires more than 24 months. If the patient really requires extraction, you may decide to, you know, meet in the middle, uh, come up with a plan that is more, uh, addressing the patient's chief complaint. So I think, uh, I mean, I'm really excited about the future possibilities of, uh, AI, uh, in, in orthodontics.

You know, the potential is tremendous.

Amy Epstein: So when we first brought you on, I mean, yes, truly, because, uh, imagine how precise you could get with all of those different variables and inputs and, uh, how much more effective treatment would be, how much shorter treatment would be, how many fewer issues we would have.

Dr. Shankar Rengasamy Venugopalan: Absolutely. You know, it's anytime we think about a technology, right? If you don't understand its limitation, it could be, it could lead us into a direction that might not be fruitful for any of us. Right? So the 1st, um, big limitation is, um, from the context of our own research. So we We conducted a study when I was in the University of Iowa, the goal really was to take a pantomime graph and use that pantomime graph taken, let's say, right around the age of nine, eight, to predict whether the erupting canine is on a path towards a normal eruption or Is there a [00:22:00] risk for an impaction or impeded eruption, right?

So an pantomograph is taken, will look at the angulation of the canine, the amount of space available, uh, whether there is, uh, hyperplasia, like, you know, the jaw size is small, factor all that in, and make a decision, maybe we should intervene now, or maybe, Let's wait. So from that point of view, what we decided to do was we took a whole bunch of pantomographs of, of patients who had an outcome of an impacted canine.

So you have a sequence of images. Then what we did was we labeled the images in the experimental group as impacted canine, and we label the control group as normal canine. And then we used a, a program that was originally designed to identify, uh, the types of bird. If you took a picture of a bird, and you give it to the algorithm, and you don't know what that bird is, the algorithm will say this The picture that you submitted resembles more like a red robin, for example, right?

So that's one big limitation. So if you think about it, I've developed an AI model system for an application in orthodontics, and then it's not be all and end all. So there has to be a way where you're constantly training this model to keep its accuracy really high. If that part of it was not taken care of, it could be leading us into an erroneous conclusion, and Unnecessary treatments or even missed opportunities, right?

This pantomime graph. will have a canine that is going to erupt normally, right? Now, when you ask an orthodontist, how did you come to that decision? The orthodontist will say, well, there is enough space, um, in the arch for that canine to erupt. The angulation of the canine is straight. Therefore, uh, the canine is going to be on the normal path of eruption.

So, okay, so we, we have trained a model system where it is making correct predictions 70 percent of the time, let's say, or 80 percent of the time, or even 90 percent of the time. It is important that the end user must know how did the algorithm Come up with that decision, right? So if you don't know that aspect of it, and it becomes really a challenge when the algorithm is marketed by a company, it becomes a proprietary uh, aspect of that company.

They may not share that information. So which means then the company has to at least provide some type of a data to convince you that it has very high levels of accuracy, and then they are constantly training the model system to stay. Relevant and accurate.

Um, it's clear to me like software AI software to determine cephalometric points or CVM are. Good tools to help orthodontists make good decisions. And the data to collect that seems to be pretty straightforward. But now when we, we talk about treatment planning, for example, now it becomes a little cloudier and, um, the more data that you have, the purity of your result is going to be, and currently I would say that the leading aligner company.

But based on our data, if you click here and make these changes to your treatment plan, you will have a 96 percent chance of success. So almost like, uh, uh, eliminating the orthodontist in terms of the, the The most important aspect of our job, which is the the diagnosis and the treatment plan. Am I off base on this?

Or is this a concern?

It's the, the doctor who has established patient doctor relationship with that patient, right? So from that sense, the AI could be making predictions or AI could be making suggestions. And I don't think we have to be worried about in the context of AI is going to be replacing the. Uh, orthodontist. So at the end of the day, the way the tooth moves, there is a biological component associated with it.

Um, Will it, uh, would, will it ever completely replace the orthodontist and you can go and sell stuff directly to, uh, to the patients? And I don't think our current, uh, legal system is structured that way. So the liability has to be assumed by someone, right? You cannot go and sue, uh, an algorithm if things went, didn't go well.

And then it turned out to be that was not the case. Right, you, you very much needed an orthodontist to finish, uh, to an optimal, uh, occlusion. And then came, uh, the aligner, uh, therapy. And everybody was worried that, you know, I'm going to print this plastic material, put it on, it's going to move teeth, and the orthodontist will be eliminated in the picture.

It's more to do with the inherent limitation of The tool itself, when you're using plastic to move teeth versus the brackets and wires, which have evolved over many decades, has a much better, um, um, predictable treatment outcome as opposed to the inherent limitation of the plastic itself. So from that, from that, from these points of view, I think that, um, even, um, When AI is integrated into, uh, this leading Aligna company that you're talking about, or as a matter of fact, any Aligna company, you cannot completely eliminate the doctor from the picture.

Um, and we are not factoring that as part of our AI application. So, the inherent limitation of the tool, the patient's biology, and then, uh, the various other factors that we talked about, Uh, we'll, we'll, we'll keep the doctor very much. So we don't have to be worried about that. And that worry should not prevent us from embracing a technology that could help us produce a desired outcome or a good outcome in an efficient way.

And my, my response to them usually my almost always is that, um, That a I won't replace you, but your competitors that are leveraging a I in their practice just might, uh, meaning that, you know, it's important for us to be not afraid of the technology and see how we can use it to benefit our patients. I'm curious.

Have you ever gotten a resident that asked you that question? And and how did you respond to it

Dr. Shankar Rengasamy Venugopalan: all the time? Um, So, you know, interestingly, orthodontics still, uh, happens to be the most, uh, soft doctor specialty. Every year we receive anywhere between, last year we received about 300 applications, and this year we received a little over 400 applications.

You

Amy Epstein: know, application after application, these people seem like they couldn't possibly exist. They're so accomplished.

Dr. Shankar Rengasamy Venugopalan: Exactly. So, um, and then when my residents ask me the, the, this is what I tell them, you know, technology, you, you, you, you cannot wish that the technology doesn't evolve. And then, you know, I use that as a safety net to protect your profession, right?

Having said that. It's, you know, me being a researcher and a clinician, I do understand what it takes to to produce, to come up with the AI model system or any type of technology to be put into use into the clinic. So there is a long runway for us to get to a point where, you know, where you, everything is fully automated.

So there was a lot of emphasis on biomechanics. Now, with the advent of TAD, some of the crazy complex movements that orthodontists thought would never be possible can be done now. So which means these tools and gadgets are going to help make your job better. Then the question is, what, what does it mean if I have to go through an orthodontic residency program to become a specialist?

So that should be the focus. You know, when residents get hung up too much on where do I position the bracket, yeah, that is important. But what is more important is you have to think about the patient as a whole, and that's where the P4 paradigm comes into the picture.

Amy Epstein: Shankar, thank you so much for all this, uh, really enlightening information.

Um, you know, uh, all the detail you provided where, where sometimes, um, we're not in the research side of things. Um, and so it's really, uh, nice to hear what's on the horizon, uh, your perspective on it, what you're hearing from residents, what you're telling residents. So thank you so much for being here today.

Dr. Shankar Rengasamy Venugopalan: Well, thank you, Amy. That was, uh, this was wonderful. And, uh, thank you for the opportunity to, uh, you know, talk about this exciting stuff. And thank you, Leon, for the invitation.

Amy Epstein: So if someone might have a question for you as a follow up to the AI, as a follow up to what it's like to be a student in the ortho program at Tufts, how can they reach you?

Dr. Shankar Rengasamy Venugopalan: So my email is my first name Shankar dot the first two initials of my last name or V or as in Romeo V as in Victor at Tufts. edu Shankar dot RV at Tufts. edu.

Amy Epstein: Perfect. Thank you so much again. Appreciate you being here today and we hope to have you on again.

Dr. Shankar Rengasamy Venugopalan: Well, thank you. Thank you, Amy. Thank you, Leon.

Thank you for the opportunity.

You can visit our website at pplpractice. com.

Dr. Leon Klempner: Yep. Thanks for watching and listening. And if you don't already, you can't already tell, I mean, I enjoy doing this podcast and having knowledgeable people come on to talk about topics that are important to you. Uh, if you want to contact me directly, my email is Leon at PPL practice.

Uh, it's hosted by one of our team members, Lacey Ellis, and you can get that. Everywhere you get your podcasts as well. And most importantly, remember for forward thinking orthodontist, it has never been a better time to be an orthodontist. We are in the golden age, so take advantage of it. Bye for now.

Narrator: Thank you for tuning in to the golden age of orthodontics. Subscribe now on Apple podcasts, Spotify, or visit our website at the golden age of orthodontics. com for direct links to both the audio and video versions of this episode.