AI in Dentistry: Achieving Human-Level Accuracy and Enhancing Patient Care with Wardah Inam, Co-Founder and CEO of Overjet

[00:00:04] [SPEAKER_02]: Welcome to Think Oral

[00:00:06] [SPEAKER_00]: Where we connect the unconnected between oral and physical health.

[00:00:10] [SPEAKER_02]: I'm your host, Dr. Jonathan Levine.

[00:00:13] [SPEAKER_00]: And I'm your host, Maria Filippova.

[00:00:15] [SPEAKER_02]: Let's get at it!

[00:00:21] [SPEAKER_00]: Hello everyone and welcome to another episode of the Think Oral Health podcast where myself,

[00:00:27] [SPEAKER_00]: Maria Filippova and Dr. Jonathan Levine explore the connections between oral health

[00:00:33] [SPEAKER_00]: and overall health.

[00:00:34] [SPEAKER_00]: I couldn't be more pleased to welcome our guests, our conversation partner for today.

[00:00:39] [SPEAKER_00]: And I couldn't be more pleased to say hi to my partner in crime and co-host Dr. Jonathan Levine.

[00:00:44] [SPEAKER_00]: Hello, Jonathan and hello, Wardah.

[00:00:46] [SPEAKER_02]: Hello, Maria.

[00:00:47] [SPEAKER_00]: Hi, Maria. Thank you for having me here.

[00:00:49] [SPEAKER_00]: Hi. All right. I've been excited about our topic today.

[00:00:53] [SPEAKER_00]: Maybe I'll do a quick intro for our guest today.

[00:00:57] [SPEAKER_00]: Dr. Wardah Inam is the Co-Founder and CEO of Overjet.

[00:01:01] [SPEAKER_00]: This is the leading provider of AI technology for the dental industry to help improve patient care.

[00:01:07] [SPEAKER_00]: Outwarder has background in product development.

[00:01:11] [SPEAKER_00]: She's what we like to call serial and the premier having led product development

[00:01:15] [SPEAKER_00]: and in other healthcare startup before.

[00:01:18] [SPEAKER_00]: Her expertise is biomedical imaging.

[00:01:21] [SPEAKER_00]: She earned her postdoctor.

[00:01:22] [SPEAKER_00]: She was a postdoctor fellow at MIT, the Computer Science and Artificial Intelligence Lab,

[00:01:27] [SPEAKER_00]: where her research really advanced the field in biomedical sensing,

[00:01:34] [SPEAKER_00]: using machine learning on wireless signals.

[00:01:37] [SPEAKER_00]: She's very well recognized and earned multiple awards,

[00:01:40] [SPEAKER_00]: including for her work as her PhD showed from MIT,

[00:01:46] [SPEAKER_00]: where she developed AI-powered microgrid technology.

[00:01:50] [SPEAKER_00]: I could go on and on, but we're so excited to have Wardah with us today

[00:01:54] [SPEAKER_00]: and to talk about AI and the exciting developments of AI in dentistry.

[00:01:59] [SPEAKER_00]: So Wardah, welcome.

[00:02:01] [SPEAKER_00]: Thank you.

[00:02:02] [SPEAKER_00]: So maybe I'll kick it off.

[00:02:04] [SPEAKER_00]: So you're running this phenomenal cutting-edge research

[00:02:08] [SPEAKER_00]: in biomedical imaging in wireless signals.

[00:02:13] [SPEAKER_00]: Walk us through the path or the journey from a postdoctor fellow at MIT

[00:02:20] [SPEAKER_00]: to a healthcare startup to oral health.

[00:02:22] [SPEAKER_00]: How did that confluence of cascading miracles happen?

[00:02:27] [SPEAKER_01]: So for me, I like to solve important problems

[00:02:31] [SPEAKER_01]: or what I feel are important problems to solve.

[00:02:33] [SPEAKER_01]: And that's how I've been driven since very early age.

[00:02:38] [SPEAKER_01]: And in this case, I was working on MRI data

[00:02:41] [SPEAKER_01]: and making that more quantitative faster to acquire.

[00:02:44] [SPEAKER_01]: And that's when I changed my dentist

[00:02:46] [SPEAKER_01]: and got a new treatment plan from the new dentist,

[00:02:50] [SPEAKER_01]: and which was very different than what I had received before.

[00:02:53] [SPEAKER_01]: And that was where I got interested in dental diagnosis,

[00:02:57] [SPEAKER_01]: where I asked for my x-ray, started reading dental 101,

[00:03:00] [SPEAKER_01]: anything I could find online.

[00:03:02] [SPEAKER_01]: And from that, I realized that this was a first fascinating field

[00:03:06] [SPEAKER_01]: in terms of how complicated things were,

[00:03:09] [SPEAKER_01]: what the technology that was being used

[00:03:10] [SPEAKER_01]: and how difficult then this job was

[00:03:13] [SPEAKER_01]: by using this technology, being able to provide diagnosis.

[00:03:16] [SPEAKER_01]: But then more importantly, to me, it felt like a societal problem.

[00:03:20] [SPEAKER_01]: Where more than just the problem I had faced,

[00:03:23] [SPEAKER_01]: there were people who didn't have access to dental care.

[00:03:26] [SPEAKER_01]: There were people who weren't getting access to dental care

[00:03:27] [SPEAKER_01]: but did not trust the dentist in some ways

[00:03:29] [SPEAKER_01]: and needed more information and understanding.

[00:03:32] [SPEAKER_01]: There were policies being defined

[00:03:34] [SPEAKER_01]: that could not take in value-based care

[00:03:36] [SPEAKER_01]: or other initiatives in mind.

[00:03:37] [SPEAKER_01]: So that's how for me it was complicated enough

[00:03:40] [SPEAKER_01]: that I felt that I could be involved with it for a long time

[00:03:44] [SPEAKER_01]: because I feel anything you do, you want to do it

[00:03:46] [SPEAKER_01]: for at least some amount of time to be able to have the impact,

[00:03:50] [SPEAKER_01]: as well as you could actually have really impactful outcomes here

[00:03:55] [SPEAKER_01]: because where we were starting off from, there was a lot to be done.

[00:03:58] [SPEAKER_00]: So it's a personal story, right?

[00:04:00] [SPEAKER_00]: You go into the dentist giving you a completely different treatment plan

[00:04:03] [SPEAKER_00]: for the same mouth and the same patient.

[00:04:07] [SPEAKER_00]: And so when did you know that you're onto something?

[00:04:11] [SPEAKER_00]: When did this become like from a curious rabbit hole

[00:04:14] [SPEAKER_00]: of why this is happening to me as a patient?

[00:04:16] [SPEAKER_00]: To wow, this could be a business opportunity here.

[00:04:19] [SPEAKER_01]: Oh, I think for me, I actually...

[00:04:21] [SPEAKER_01]: Well, once I got more obsessed with dentistry,

[00:04:24] [SPEAKER_01]: what that means is I was really everything I could find.

[00:04:26] [SPEAKER_01]: I was watching everything I could find.

[00:04:28] [SPEAKER_01]: I wanted to just shadow dentists.

[00:04:29] [SPEAKER_01]: That's what I wanted to do.

[00:04:32] [SPEAKER_01]: I quit my job and not knowing that this could be a business

[00:04:35] [SPEAKER_01]: or what I was doing, I started shadowing dentists

[00:04:38] [SPEAKER_01]: and trying to meet as many dentists I could

[00:04:40] [SPEAKER_01]: to figure out exactly what we were doing.

[00:04:42] [SPEAKER_01]: I don't think there was like one moment that was like,

[00:04:45] [SPEAKER_01]: oh, we made it or if there's something

[00:04:46] [SPEAKER_01]: that this is going to solve.

[00:04:48] [SPEAKER_01]: For me, it's been this continuous journey of like,

[00:04:51] [SPEAKER_01]: I was first curious then more involved

[00:04:53] [SPEAKER_01]: in trying to figure out what we could do

[00:04:55] [SPEAKER_01]: to help improve patient outcomes.

[00:04:57] [SPEAKER_01]: And then it became a project and then it became bigger

[00:05:00] [SPEAKER_01]: and bigger.

[00:05:01] [SPEAKER_01]: So I don't think there was one moment where I felt like,

[00:05:03] [SPEAKER_01]: oh, we're onto something.

[00:05:05] [SPEAKER_01]: Interesting.

[00:05:06] [SPEAKER_00]: I like the word that you just obsessed.

[00:05:08] [SPEAKER_00]: That's a very common theme, John,

[00:05:10] [SPEAKER_00]: and between all our entrepreneurs.

[00:05:13] [SPEAKER_00]: When was the last time you were obsessed about something?

[00:05:16] [SPEAKER_00]: Some of your projects, that's what really gets you going.

[00:05:19] [SPEAKER_02]: That's right.

[00:05:19] [SPEAKER_02]: The obsession and tenacity merging with curiosity

[00:05:23] [SPEAKER_02]: and throwing a big dose of humility

[00:05:27] [SPEAKER_02]: and a high level of competence

[00:05:29] [SPEAKER_02]: and you start having an entrepreneur

[00:05:31] [SPEAKER_02]: that's going to move a needle.

[00:05:33] [SPEAKER_02]: So let me ask you, you went from an idea

[00:05:36] [SPEAKER_02]: and really solving what you saw as a major problem.

[00:05:40] [SPEAKER_02]: Which as a clinician and as an academic myself,

[00:05:44] [SPEAKER_02]: realizing that if we asked five dentists looking at a full series

[00:05:48] [SPEAKER_02]: of x-rays, you might get a number of different diagnosis.

[00:05:54] [SPEAKER_02]: What did you say to yourself?

[00:05:56] [SPEAKER_02]: Knowing what I know about technology

[00:05:58] [SPEAKER_02]: and let's call it let's just say it's the new technology

[00:06:01] [SPEAKER_02]: and machine learning and AI,

[00:06:03] [SPEAKER_02]: where did you see the opportunity?

[00:06:05] [SPEAKER_02]: Unpack that for us.

[00:06:07] [SPEAKER_01]: This was an interesting time.

[00:06:08] [SPEAKER_01]: This is basically 2018

[00:06:10] [SPEAKER_01]: where you've had the introduction of Transformers

[00:06:13] [SPEAKER_01]: and you have computer vision advancement happening.

[00:06:17] [SPEAKER_01]: And I would say in computer vision in AI,

[00:06:20] [SPEAKER_01]: I think it saw the first development

[00:06:22] [SPEAKER_01]: to the point that we were able to get at the level of

[00:06:25] [SPEAKER_01]: a start to exceed dentist level or human level accuracies.

[00:06:29] [SPEAKER_01]: And because of that,

[00:06:30] [SPEAKER_01]: so there was a very interesting time

[00:06:31] [SPEAKER_01]: in terms of technology that was there

[00:06:33] [SPEAKER_01]: to actually help humans.

[00:06:35] [SPEAKER_01]: Because before that,

[00:06:36] [SPEAKER_01]: if you've had previous say 10 years ago or so,

[00:06:39] [SPEAKER_01]: the technology just wasn't there.

[00:06:41] [SPEAKER_01]: You could work on AI,

[00:06:42] [SPEAKER_01]: but it was a research project.

[00:06:45] [SPEAKER_01]: It wasn't actually impactful enough in the real world.

[00:06:48] [SPEAKER_01]: So the technology was there.

[00:06:49] [SPEAKER_01]: Also, the second piece was I watched a documentary,

[00:06:52] [SPEAKER_01]: it was called Dollars and Dentists.

[00:06:54] [SPEAKER_01]: And it talked about the entry of private equity

[00:06:57] [SPEAKER_01]: in dentistry and how the dental landscape was changing.

[00:07:02] [SPEAKER_01]: So those two things,

[00:07:03] [SPEAKER_01]: I think really played a role in my thinking of like,

[00:07:05] [SPEAKER_01]: why it should be now rather than later,

[00:07:07] [SPEAKER_01]: which was, hey, the technology is advanced enough.

[00:07:10] [SPEAKER_01]: Also the industry,

[00:07:12] [SPEAKER_01]: if we do not introduce ways of measuring quality of care,

[00:07:15] [SPEAKER_01]: measuring outcomes,

[00:07:16] [SPEAKER_01]: measuring clinical care,

[00:07:18] [SPEAKER_01]: people will not have the tools to actually build

[00:07:21] [SPEAKER_01]: organizations at scale

[00:07:22] [SPEAKER_01]: and really actually promote quality of care as well.

[00:07:25] [SPEAKER_01]: So I think those were the two aspects combining,

[00:07:28] [SPEAKER_01]: which led me to say,

[00:07:29] [SPEAKER_01]: hey, it needs to be now

[00:07:31] [SPEAKER_01]: and I need to leave my job to do this.

[00:07:33] [SPEAKER_00]: And Warda, you used an AI theorem around transformers.

[00:07:38] [SPEAKER_00]: So for our listeners,

[00:07:39] [SPEAKER_00]: could you explain why transformers are as part of the

[00:07:42] [SPEAKER_00]: natural language processing models

[00:07:45] [SPEAKER_00]: are better suited to tackle some of the problems

[00:07:48] [SPEAKER_00]: that dentistry has versus other technology and other AI?

[00:07:52] [SPEAKER_00]: Because a lot of people think about AI

[00:07:54] [SPEAKER_00]: as this monolithic thing,

[00:07:55] [SPEAKER_00]: but there's multiple models that are underneath that category

[00:08:00] [SPEAKER_00]: and transformers was an important part

[00:08:02] [SPEAKER_00]: in terms of the technology evolution.

[00:08:04] [SPEAKER_00]: So why were they better suited than other

[00:08:07] [SPEAKER_00]: naturally language processing models, for example?

[00:08:10] [SPEAKER_01]: I think that this was research that happened.

[00:08:13] [SPEAKER_01]: Then there's background behind

[00:08:14] [SPEAKER_01]: before the transformer model got created as well.

[00:08:17] [SPEAKER_01]: It happened at Google

[00:08:18] [SPEAKER_01]: and Ashish Sani was leading this effort

[00:08:22] [SPEAKER_01]: and it basically, I would say,

[00:08:25] [SPEAKER_01]: the paper itself was called attention is all you need.

[00:08:27] [SPEAKER_01]: So it was around how the attention mechanism leads

[00:08:30] [SPEAKER_01]: to really great outputs,

[00:08:32] [SPEAKER_01]: but also they really simplified the architecture as well.

[00:08:35] [SPEAKER_01]: And I think what that ended up doing

[00:08:37] [SPEAKER_01]: was like having similar people could now,

[00:08:40] [SPEAKER_01]: you could be in computer vision research

[00:08:42] [SPEAKER_01]: and do natural language processing as well.

[00:08:45] [SPEAKER_01]: So these were divergent fields before

[00:08:47] [SPEAKER_01]: and now you actually started having models

[00:08:50] [SPEAKER_01]: that you could utilize both,

[00:08:52] [SPEAKER_01]: which we actually utilize.

[00:08:53] [SPEAKER_01]: We actually do computer vision and natural language processing.

[00:08:57] [SPEAKER_01]: So that was one, but also it was just,

[00:09:01] [SPEAKER_01]: we could also have more context.

[00:09:03] [SPEAKER_01]: You weren't learning without context as well.

[00:09:06] [SPEAKER_01]: So that was one of the things that I think

[00:09:07] [SPEAKER_01]: it just ended up getting to a point

[00:09:10] [SPEAKER_01]: that in the end what we cared about was accuracy,

[00:09:13] [SPEAKER_01]: like where the accuracy was improving significantly

[00:09:17] [SPEAKER_01]: by these models.

[00:09:18] [SPEAKER_01]: And also I think the one more fundamental thing

[00:09:20] [SPEAKER_01]: about transform models was the fact

[00:09:22] [SPEAKER_01]: that they really utilize the GPU architecture effectively.

[00:09:28] [SPEAKER_01]: So you could parallelize a lot of things.

[00:09:30] [SPEAKER_01]: So you could again get really good outputs effectively as well.

[00:09:34] [SPEAKER_00]: That's right.

[00:09:34] [SPEAKER_00]: So faster compute convergent models

[00:09:37] [SPEAKER_00]: coming up with more accurate,

[00:09:40] [SPEAKER_00]: quote unquote accurate model.

[00:09:41] [SPEAKER_00]: So I'm going to channel Dr. Levine here.

[00:09:44] [SPEAKER_00]: How do you measure accuracy and quality

[00:09:47] [SPEAKER_00]: in care and also in models, in AI models?

[00:09:51] [SPEAKER_01]: So I think there's one what happens in model development

[00:09:56] [SPEAKER_01]: and then there's what happens in, for example,

[00:09:58] [SPEAKER_01]: if you're getting a clear, the clinical studies.

[00:10:01] [SPEAKER_01]: So during model development,

[00:10:02] [SPEAKER_01]: there's the sensitivity specificity metrics

[00:10:05] [SPEAKER_01]: that are being utilized to actually measure accuracy

[00:10:08] [SPEAKER_01]: of the models.

[00:10:09] [SPEAKER_01]: However, the way we do it is first,

[00:10:11] [SPEAKER_01]: we get to the right metrics that we've set for ourselves

[00:10:14] [SPEAKER_01]: which we determine by what is human level accuracy.

[00:10:17] [SPEAKER_01]: Human level accuracy is not 100% as we all know.

[00:10:20] [SPEAKER_01]: So for every output, there might be a different human level accuracy.

[00:10:24] [SPEAKER_01]: Say two numbers, people are very good at determining

[00:10:27] [SPEAKER_01]: what two numbers are.

[00:10:28] [SPEAKER_01]: Carey's lesion is probably not as good as two numbers.

[00:10:31] [SPEAKER_01]: So that means there's different accuracy levels

[00:10:33] [SPEAKER_01]: on what two numbers are and what carries are.

[00:10:36] [SPEAKER_01]: First you set those and then you say,

[00:10:38] [SPEAKER_01]: we need to reach or surpass this accuracy.

[00:10:41] [SPEAKER_01]: And then once it's done models

[00:10:43] [SPEAKER_01]: or the metrics are achieved,

[00:10:45] [SPEAKER_01]: you might have two outputs

[00:10:47] [SPEAKER_01]: or two models where both have achieved that.

[00:10:50] [SPEAKER_01]: Then we actually give it to clinicians

[00:10:52] [SPEAKER_01]: and say, okay, which one is better?

[00:10:54] [SPEAKER_01]: So like having a subjective opinion on it as well

[00:10:57] [SPEAKER_01]: and getting that maybe one is making mistakes

[00:10:59] [SPEAKER_01]: on things that you should never make a mistake on

[00:11:02] [SPEAKER_01]: and the other is making mistakes

[00:11:03] [SPEAKER_01]: which people will tolerate.

[00:11:04] [SPEAKER_01]: So having that subjective opinion is helpful as well.

[00:11:07] [SPEAKER_01]: And then so that happens during training.

[00:11:09] [SPEAKER_01]: And then what we do is we have to actually pass

[00:11:13] [SPEAKER_01]: FDA clearances.

[00:11:15] [SPEAKER_01]: And what the FDA's metric or bar is not only

[00:11:18] [SPEAKER_01]: that it needs to be as good as humans,

[00:11:20] [SPEAKER_01]: but their bar is that when humans utilize it

[00:11:23] [SPEAKER_01]: or dentists utilize it, they need to be better.

[00:11:25] [SPEAKER_01]: So what you need to do is then create a ground truth set

[00:11:28] [SPEAKER_01]: of what actually is the ground truth.

[00:11:30] [SPEAKER_01]: And then you need to make sure that you're actually showing

[00:11:33] [SPEAKER_01]: that not only are these models better,

[00:11:35] [SPEAKER_01]: but when humans are utilizing it, they're better as well.

[00:11:38] [SPEAKER_01]: And that's the again, the second part of accuracy

[00:11:40] [SPEAKER_01]: to where you need to, and if they're not getting better,

[00:11:43] [SPEAKER_01]: then that's a problem.

[00:11:44] [SPEAKER_01]: You don't want to put that into market.

[00:11:47] [SPEAKER_01]: And for us, we do not push anything into production

[00:11:50] [SPEAKER_01]: or into our customers until it has surpassed

[00:11:55] [SPEAKER_01]: that human level capabilities.

[00:11:57] [SPEAKER_02]: Let's go there a little bit.

[00:11:58] [SPEAKER_02]: So we know about healthcare.

[00:12:01] [SPEAKER_02]: We know about adoption rates and healthcare, the average.

[00:12:05] [SPEAKER_02]: Maria, we always talk about takes about 17 years

[00:12:08] [SPEAKER_02]: for adoption dentistry is probably a little bit longer

[00:12:10] [SPEAKER_02]: when we look at the digital workflow

[00:12:12] [SPEAKER_02]: and the different technologies we have today in dentistry.

[00:12:16] [SPEAKER_02]: What do you find being for adoption with overjet

[00:12:21] [SPEAKER_02]: and with the AI driven model that is so relevant

[00:12:25] [SPEAKER_02]: and amazing how we can answer these big issues

[00:12:29] [SPEAKER_02]: in dentistry of subjectivity and validation

[00:12:33] [SPEAKER_02]: and quality of care?

[00:12:35] [SPEAKER_02]: What are you seeing from an adoption standpoint

[00:12:37] [SPEAKER_02]: on the professional side?

[00:12:40] [SPEAKER_01]: So I think adoption of technology is just accelerating,

[00:12:44] [SPEAKER_01]: whether it's in dentistry or any other field.

[00:12:47] [SPEAKER_01]: How fast things used to move,

[00:12:49] [SPEAKER_01]: that's not how fast things are moving.

[00:12:51] [SPEAKER_01]: We can all see the example of chat GPD,

[00:12:53] [SPEAKER_01]: imagine a product coming into the market

[00:12:56] [SPEAKER_01]: and becoming such or being used by so many people this quickly.

[00:13:01] [SPEAKER_01]: So I think that's where, and I do think,

[00:13:03] [SPEAKER_01]: of course, like digital extras was amazing technology,

[00:13:06] [SPEAKER_01]: but it was a hardware technology.

[00:13:07] [SPEAKER_01]: People had to make these big capital investments

[00:13:10] [SPEAKER_01]: to change the product.

[00:13:12] [SPEAKER_01]: So first, if it's software technology,

[00:13:14] [SPEAKER_01]: then on top of it, you have just the pace of innovation

[00:13:17] [SPEAKER_01]: happening, people are more connected than ever.

[00:13:19] [SPEAKER_01]: There's social media, there is Facebook ads,

[00:13:22] [SPEAKER_01]: there's these conferences that are happening

[00:13:23] [SPEAKER_01]: and just growing in number.

[00:13:25] [SPEAKER_01]: So I think adoption is happening faster.

[00:13:27] [SPEAKER_01]: So for us, we have two customer segments.

[00:13:30] [SPEAKER_01]: We sell to insurance companies

[00:13:31] [SPEAKER_01]: as well as we sell to dental practices.

[00:13:33] [SPEAKER_01]: On the insurance company side,

[00:13:35] [SPEAKER_01]: now we serve 110 million members.

[00:13:36] [SPEAKER_01]: So that is about 50% adoption, if you consider it that way.

[00:13:41] [SPEAKER_01]: And then on the practice side,

[00:13:43] [SPEAKER_01]: I would say it's hard to sell one by one to every group,

[00:13:47] [SPEAKER_01]: but like we have about 100 plus DSOs that we serve.

[00:13:51] [SPEAKER_01]: And then we have the smaller segment

[00:13:53] [SPEAKER_01]: of solo practices as well.

[00:13:55] [SPEAKER_01]: So I do think it's probably, we don't have a curve.

[00:13:58] [SPEAKER_01]: I hope somebody puts a curve,

[00:14:00] [SPEAKER_01]: but probably it's one of the fastest adopting

[00:14:03] [SPEAKER_01]: adopted technologies in dentistry.

[00:14:06] [SPEAKER_02]: Yeah. And when you get into these dental practices,

[00:14:10] [SPEAKER_02]: even DSOs and the singular type practices,

[00:14:12] [SPEAKER_02]: to get the staff to utilize it day in and day out.

[00:14:16] [SPEAKER_02]: So it's in the computer systems,

[00:14:19] [SPEAKER_02]: the patients are coming through their normal workflow.

[00:14:23] [SPEAKER_02]: What have you found during the team,

[00:14:25] [SPEAKER_02]: the training team and the people in your organization

[00:14:28] [SPEAKER_02]: that are focusing on that?

[00:14:31] [SPEAKER_02]: How do you find the adoption of the actual,

[00:14:33] [SPEAKER_02]: the practice because dentistry is so different, right?

[00:14:35] [SPEAKER_02]: It's not hospital based.

[00:14:37] [SPEAKER_02]: And even the DSOs, most of them are allowing the dental practices

[00:14:40] [SPEAKER_02]: to have a certain level of clinical autonomy.

[00:14:43] [SPEAKER_02]: So how have you found getting these dental practices

[00:14:47] [SPEAKER_02]: to people that hygienists the assistants,

[00:14:49] [SPEAKER_02]: in addition to the doctors, let's say wanting to integrate it?

[00:14:53] [SPEAKER_02]: How do you see that happening in actuality?

[00:14:56] [SPEAKER_02]: And do you have certain measurements or KPIs against that?

[00:14:59] [SPEAKER_02]: And how do you see the future rolling out for that?

[00:15:02] [SPEAKER_01]: Yeah. And so I know that I think that's an extremely important question

[00:15:07] [SPEAKER_01]: that you asked and probably you've hit the nail in the head around

[00:15:11] [SPEAKER_01]: the challenges in adoption,

[00:15:13] [SPEAKER_01]: which isn't somebody buying the technology.

[00:15:16] [SPEAKER_01]: It's actually the change management that needs to happen

[00:15:18] [SPEAKER_01]: in these practices for wide scale adoption

[00:15:22] [SPEAKER_01]: or significant adoption in these practices as well.

[00:15:25] [SPEAKER_01]: So I think in this case for us,

[00:15:28] [SPEAKER_01]: what has been very important is first the training piece of it,

[00:15:32] [SPEAKER_01]: which is very hard if you think about it

[00:15:33] [SPEAKER_01]: compared to a normal software where you have one person to train.

[00:15:37] [SPEAKER_01]: In the dental practice, you have to train the entire staff.

[00:15:40] [SPEAKER_01]: And when you're trying to train the entire staff,

[00:15:42] [SPEAKER_01]: that means that they need to be present as well.

[00:15:45] [SPEAKER_01]: That means everybody needs to be present

[00:15:46] [SPEAKER_01]: in that the day that training is happening.

[00:15:49] [SPEAKER_01]: Then there's a lot of turnover that happens in practices as well.

[00:15:53] [SPEAKER_01]: That means new staff is coming in,

[00:15:55] [SPEAKER_01]: you're training the new staff as well.

[00:15:57] [SPEAKER_01]: So which is something that we measure pretty effectively

[00:16:00] [SPEAKER_01]: or try to measure significantly as well.

[00:16:02] [SPEAKER_01]: And I think for us, the training aspect is number one.

[00:16:06] [SPEAKER_01]: Then number two is the champion buy-in.

[00:16:10] [SPEAKER_01]: So who is the person who is responsible for that adoption?

[00:16:13] [SPEAKER_01]: And it actually varies in different practices.

[00:16:15] [SPEAKER_01]: It's different.

[00:16:16] [SPEAKER_01]: Sometimes you have a dentist who is very prescriptive

[00:16:20] [SPEAKER_01]: on what needs to happen in the practice,

[00:16:21] [SPEAKER_01]: but that's not always the case.

[00:16:23] [SPEAKER_01]: Then you have the office manager who's leading

[00:16:25] [SPEAKER_01]: some of the practices and what gets adopted.

[00:16:27] [SPEAKER_01]: And then in some others, it might even be

[00:16:30] [SPEAKER_01]: if the hygienist has been there for a while,

[00:16:32] [SPEAKER_01]: they even might have a lot of saying

[00:16:33] [SPEAKER_01]: what gets adopted in the practices.

[00:16:36] [SPEAKER_01]: So figuring out who that person is in the practice

[00:16:38] [SPEAKER_01]: and identifying that person is very important as well.

[00:16:42] [SPEAKER_01]: And then the third is utilizing data

[00:16:44] [SPEAKER_01]: to understand what is happening across the practices.

[00:16:47] [SPEAKER_01]: So we do measure what we call the number of actions

[00:16:50] [SPEAKER_01]: in the practice.

[00:16:52] [SPEAKER_01]: Then we have the daily active users.

[00:16:54] [SPEAKER_01]: So are people logging in and utilizing the product?

[00:16:57] [SPEAKER_01]: And then you have the providing the tools to, say,

[00:17:00] [SPEAKER_01]: the SO to ensure that it is being utilized

[00:17:03] [SPEAKER_01]: across their practices as well.

[00:17:04] [SPEAKER_02]: Right. That's so interesting.

[00:17:06] [SPEAKER_02]: Maria, there's so much fragmented dentistry industry

[00:17:09] [SPEAKER_02]: starting to have a business model

[00:17:11] [SPEAKER_02]: that's growing up and getting more organized.

[00:17:14] [SPEAKER_02]: There's so much personal bias

[00:17:16] [SPEAKER_02]: with these individual practice owners

[00:17:18] [SPEAKER_02]: or these different type of practices.

[00:17:21] [SPEAKER_02]: It seems to me if I would offer up

[00:17:23] [SPEAKER_02]: some of my own insight of being a clinician

[00:17:26] [SPEAKER_02]: that the AI reading x-rays needs to be automatic.

[00:17:30] [SPEAKER_02]: You take an x-ray, you have an automatic read.

[00:17:33] [SPEAKER_02]: And almost the subjectivity of yes,

[00:17:35] [SPEAKER_02]: let's go put some AI on this x-ray, vaporize this.

[00:17:39] [SPEAKER_02]: Pans, x-ray, digital x-rays,

[00:17:42] [SPEAKER_02]: automatically read by the software.

[00:17:45] [SPEAKER_02]: And that number gets right up in front.

[00:17:48] [SPEAKER_02]: And I think when the offices choose to do that,

[00:17:51] [SPEAKER_02]: I think that's going to be a good turning point.

[00:17:53] [SPEAKER_02]: And we're getting there.

[00:17:54] [SPEAKER_02]: I remember the early days, three, four, five years ago

[00:17:56] [SPEAKER_02]: of having AI reading x-rays

[00:17:59] [SPEAKER_02]: and hearing some of the comments from colleagues

[00:18:02] [SPEAKER_02]: of more of a fear-based conversation

[00:18:04] [SPEAKER_02]: than of optimistic positive one

[00:18:07] [SPEAKER_02]: of really having an AI co-pilot

[00:18:10] [SPEAKER_02]: helping you diagnose

[00:18:12] [SPEAKER_02]: where I think we're finally getting to today.

[00:18:15] [SPEAKER_02]: Would you say that's fair?

[00:18:17] [SPEAKER_00]: I think it's fair.

[00:18:18] [SPEAKER_00]: I mean overcoming the inertia

[00:18:19] [SPEAKER_00]: of we've always done it this way.

[00:18:22] [SPEAKER_00]: It's a big barrier.

[00:18:24] [SPEAKER_00]: And also going back to dollars and dentistry,

[00:18:28] [SPEAKER_00]: a lot of these practices are fairly successful.

[00:18:32] [SPEAKER_00]: So if it's not broken, why fix it?

[00:18:34] [SPEAKER_00]: I don't have a problem.

[00:18:36] [SPEAKER_00]: Why do I have AI?

[00:18:37] [SPEAKER_00]: I'm doing perfectly okay with my treatment plans.

[00:18:40] [SPEAKER_00]: No one's come to complain.

[00:18:41] [SPEAKER_00]: And so from an individual's sole practice owner,

[00:18:45] [SPEAKER_00]: the incentive to change

[00:18:47] [SPEAKER_00]: or to try something new is minimal.

[00:18:50] [SPEAKER_00]: And so that's the uphill battle that...

[00:18:53] [SPEAKER_00]: And so I think slowly kicking and screaming,

[00:18:57] [SPEAKER_00]: a lot of these dental practices are getting moved into

[00:19:01] [SPEAKER_00]: the future of dentistry

[00:19:02] [SPEAKER_00]: where dentistry is augmented by technology

[00:19:06] [SPEAKER_00]: rather than replaced by technology.

[00:19:09] [SPEAKER_00]: And maybe if we pan in on that future

[00:19:12] [SPEAKER_00]: of augmented dental care,

[00:19:15] [SPEAKER_00]: where do you see AI evolving?

[00:19:18] [SPEAKER_00]: Do we see at some point where certain procedures

[00:19:22] [SPEAKER_00]: or activities in the dental office,

[00:19:24] [SPEAKER_00]: if they're not done with AI first,

[00:19:28] [SPEAKER_00]: they're not done up to the standard of care?

[00:19:31] [SPEAKER_00]: When does AI become the default

[00:19:34] [SPEAKER_00]: for certain procedures?

[00:19:36] [SPEAKER_00]: Or is that at all the path we wanted to go?

[00:19:39] [SPEAKER_01]: I was just looking up what our numbers looked like

[00:19:42] [SPEAKER_01]: on the question around adoption.

[00:19:44] [SPEAKER_01]: So I think if you look at any other AI technology

[00:19:47] [SPEAKER_01]: being utilized in healthcare,

[00:19:49] [SPEAKER_01]: I think our numbers are much higher than that.

[00:19:51] [SPEAKER_01]: So for us it is 90% weekly active clinic usage,

[00:19:56] [SPEAKER_01]: 80% daily active clinic usage.

[00:19:58] [SPEAKER_01]: That means every...

[00:20:00] [SPEAKER_01]: And these are DSOs.

[00:20:01] [SPEAKER_01]: So that means even if it was a top down buying decision,

[00:20:05] [SPEAKER_01]: 80% of the practices are using it daily

[00:20:07] [SPEAKER_01]: and 90% are using it at least

[00:20:10] [SPEAKER_01]: weekly might mean multiple days.

[00:20:12] [SPEAKER_01]: They haven't used it every day of the week.

[00:20:13] [SPEAKER_01]: And there are days where it's Friday, etc.

[00:20:15] [SPEAKER_01]: That might be less usage.

[00:20:17] [SPEAKER_01]: Or days where you might just be doing procedures

[00:20:19] [SPEAKER_01]: and not doing diagnostics.

[00:20:21] [SPEAKER_01]: So those are some of the things to put even more caveat on it

[00:20:24] [SPEAKER_01]: because it doesn't need to be used on every appointment as well.

[00:20:27] [SPEAKER_01]: It needs to be used when data is being collected

[00:20:29] [SPEAKER_01]: and being looked at least currently the way the technology is.

[00:20:32] [SPEAKER_01]: I think it'll change very soon.

[00:20:34] [SPEAKER_01]: But that's where I think the technology is.

[00:20:36] [SPEAKER_01]: If you look at it in context,

[00:20:38] [SPEAKER_01]: I think it's been adopted in dentistry

[00:20:40] [SPEAKER_01]: unlike anything else in healthcare.

[00:20:43] [SPEAKER_01]: So better adoption in dentistry.

[00:20:45] [SPEAKER_01]: But there is room for improvement.

[00:20:47] [SPEAKER_01]: And I think this is one where the products get more mature.

[00:20:50] [SPEAKER_01]: We're going to see an increase in that

[00:20:52] [SPEAKER_01]: in these numbers as well.

[00:20:54] [SPEAKER_01]: Now, to your question around what does it look like

[00:20:57] [SPEAKER_01]: in the future?

[00:20:58] [SPEAKER_01]: I do think in the end, if you have an amazing dentist

[00:21:01] [SPEAKER_01]: and they are pretty confident

[00:21:04] [SPEAKER_01]: what they're doing, etc.

[00:21:06] [SPEAKER_01]: I wouldn't say it has to be used in every case, etc.

[00:21:09] [SPEAKER_01]: And that's just knowing how dentists are.

[00:21:12] [SPEAKER_01]: But it is one of those things

[00:21:13] [SPEAKER_01]: where as it becomes part of your workflow

[00:21:17] [SPEAKER_01]: and part of how you do things,

[00:21:19] [SPEAKER_01]: just people get used to it as well.

[00:21:21] [SPEAKER_01]: So it is like if you were drawing a spell check, for example,

[00:21:24] [SPEAKER_01]: when you're doing Grammarly and utilizing spell check,

[00:21:27] [SPEAKER_01]: of course you could get all the grammar

[00:21:29] [SPEAKER_01]: if somebody wanted to think through everything

[00:21:31] [SPEAKER_01]: and wanted to spell and could spell check everything.

[00:21:34] [SPEAKER_01]: However, once you start getting used to spell check

[00:21:36] [SPEAKER_01]: and Grammarly helping you crowd better sentences,

[00:21:40] [SPEAKER_01]: very quickly that becomes a default as well.

[00:21:44] [SPEAKER_01]: So I think it is one of those things

[00:21:45] [SPEAKER_01]: that as the technology improves further and further,

[00:21:48] [SPEAKER_01]: we will see especially for people who like to do an excellent job

[00:21:53] [SPEAKER_01]: to utilize this technology not basically every time

[00:21:57] [SPEAKER_01]: because it's just the way you put it.

[00:21:59] [SPEAKER_01]: It's the default.

[00:22:00] [SPEAKER_01]: But I don't think it needs to be enforced on people

[00:22:03] [SPEAKER_01]: unless it is providing them value,

[00:22:05] [SPEAKER_01]: they are going to adopt it.

[00:22:06] [SPEAKER_00]: At this point, let me ask you if you believe this is truth

[00:22:12] [SPEAKER_00]: or maybe a myth that we need to disperse.

[00:22:15] [SPEAKER_00]: Currently there is a live and thriving discussion

[00:22:18] [SPEAKER_00]: around adoption of AI in dental programs

[00:22:22] [SPEAKER_00]: where we train the future generation of dentists.

[00:22:25] [SPEAKER_00]: And there's a concern, maybe rightfully so,

[00:22:29] [SPEAKER_00]: that if dentists are not reading the images themselves

[00:22:32] [SPEAKER_00]: but AI is reading them first,

[00:22:34] [SPEAKER_00]: they're not going to be as well equipped as another dentist,

[00:22:40] [SPEAKER_00]: perhaps Jonathan Levine's cohort of dentists

[00:22:43] [SPEAKER_00]: who didn't have technology to aid them.

[00:22:46] [SPEAKER_00]: They had to go through all these images

[00:22:49] [SPEAKER_00]: and pattern recognition themselves

[00:22:50] [SPEAKER_00]: to create that sense of what's early lesion, what's not.

[00:22:54] [SPEAKER_00]: So where do you stand on that question

[00:22:57] [SPEAKER_00]: whether or not AI in training settings in dental schools

[00:23:03] [SPEAKER_00]: helps us or deters from the clinical education?

[00:23:08] [SPEAKER_00]: And that's a question for both of you.

[00:23:10] [SPEAKER_00]: But Warda, please jump in here first.

[00:23:12] [SPEAKER_01]: I said two conflicting things there.

[00:23:14] [SPEAKER_01]: So one is the answer that is based on

[00:23:18] [SPEAKER_01]: University of Florida's research.

[00:23:20] [SPEAKER_01]: University of Florida conducted the research

[00:23:23] [SPEAKER_01]: at Linden Overjet and what they did was

[00:23:25] [SPEAKER_01]: they trained people using Overjet

[00:23:27] [SPEAKER_01]: and then dentists did not have

[00:23:30] [SPEAKER_01]: or some dental students did not have Overjet utilized

[00:23:34] [SPEAKER_01]: and they looked at the accuracy, not only while

[00:23:36] [SPEAKER_01]: they were utilizing Overjet but after the fact.

[00:23:39] [SPEAKER_01]: And they found that the students who had used AI to learn

[00:23:43] [SPEAKER_01]: were much better at diagnosing

[00:23:44] [SPEAKER_01]: and these numbers are crazy.

[00:23:45] [SPEAKER_01]: They are going to publish this study.

[00:23:47] [SPEAKER_01]: So I don't want to disclose everything there.

[00:23:49] [SPEAKER_01]: And I think the Dean has presented as well

[00:23:51] [SPEAKER_01]: yet but hasn't been, I think, published yet.

[00:23:54] [SPEAKER_01]: But there is research that University of Florida did

[00:23:57] [SPEAKER_01]: around showing that the students are much better

[00:24:00] [SPEAKER_01]: after learning.

[00:24:02] [SPEAKER_01]: However, the conflicting part is intuitively, I think,

[00:24:06] [SPEAKER_01]: once people get used to it, if we look historically,

[00:24:09] [SPEAKER_01]: think about, for example, calculators.

[00:24:11] [SPEAKER_01]: Like before calculators, the way my dad can do math in his head,

[00:24:14] [SPEAKER_01]: probably I cannot do math in my head

[00:24:16] [SPEAKER_01]: because I grew up on calculators

[00:24:18] [SPEAKER_01]: and the next generation will grow up

[00:24:20] [SPEAKER_01]: with LLMs doing their math where they didn't,

[00:24:21] [SPEAKER_01]: wouldn't even have to put in the numbers

[00:24:23] [SPEAKER_01]: into a calculator.

[00:24:24] [SPEAKER_01]: And of course his math in his head was better than mine,

[00:24:27] [SPEAKER_01]: the math I could do and then

[00:24:29] [SPEAKER_01]: the math my child will be able to do.

[00:24:31] [SPEAKER_01]: However, is it better to get the better results

[00:24:34] [SPEAKER_01]: or is it better for me to put the numbers

[00:24:37] [SPEAKER_01]: and calculate it myself?

[00:24:38] [SPEAKER_01]: If we look at just how math has evolved as well,

[00:24:41] [SPEAKER_01]: of course we are doing things that were never possible before

[00:24:44] [SPEAKER_01]: and we are getting to solving problems

[00:24:46] [SPEAKER_01]: that have not been solved in a while as well.

[00:24:49] [SPEAKER_01]: So I do think just technological progress

[00:24:51] [SPEAKER_01]: is inevitable.

[00:24:52] [SPEAKER_01]: It needs to happen and we just need to evolve

[00:24:55] [SPEAKER_01]: and say, hey, now we probably,

[00:24:56] [SPEAKER_01]: this is something that a calculator

[00:24:58] [SPEAKER_01]: can do easily or say LLMs can do easily.

[00:25:01] [SPEAKER_01]: How do we get to the next level?

[00:25:03] [SPEAKER_01]: How do we educate our students to not think about the basics

[00:25:07] [SPEAKER_01]: because say the basics will be done,

[00:25:09] [SPEAKER_01]: how do we think about let's put better education

[00:25:12] [SPEAKER_01]: around edge cases?

[00:25:13] [SPEAKER_01]: Let's think about backband effects.

[00:25:15] [SPEAKER_01]: Let's think about other areas that we can improve

[00:25:17] [SPEAKER_01]: such as the group of people we do better

[00:25:19] [SPEAKER_01]: rather than us versus technology.

[00:25:22] [SPEAKER_00]: Yeah, the dentist of the future

[00:25:23] [SPEAKER_00]: needs different skillset.

[00:25:25] [SPEAKER_00]: Like recognizing what's noise,

[00:25:27] [SPEAKER_00]: what's actual AI, LLM hallucinations,

[00:25:31] [SPEAKER_00]: all these things are gonna have to be part of the skill set.

[00:25:34] [SPEAKER_00]: Jonathan, go ahead.

[00:25:35] [SPEAKER_00]: I'm keen to do what you're saying.

[00:25:36] [SPEAKER_02]: That's exactly what you just said

[00:25:38] [SPEAKER_02]: and what I wanted to say

[00:25:39] [SPEAKER_02]: and that's exactly why you have to start with the youth,

[00:25:42] [SPEAKER_02]: educating the dental schools

[00:25:44] [SPEAKER_02]: so that having that AI co-pilot,

[00:25:48] [SPEAKER_02]: learning through having that AI

[00:25:50] [SPEAKER_02]: and really understanding it

[00:25:53] [SPEAKER_02]: and then taking that out to clinical practice,

[00:25:55] [SPEAKER_02]: it just becomes part of the protocol in their system

[00:25:58] [SPEAKER_02]: and it's not surprising, in fact,

[00:26:00] [SPEAKER_02]: easy to predict that their capability skill set

[00:26:03] [SPEAKER_02]: of reading x-rays is so much higher

[00:26:05] [SPEAKER_02]: and so much better.

[00:26:06] [SPEAKER_02]: And while we're on the conversation of future,

[00:26:08] [SPEAKER_02]: why take us to the future a little bit?

[00:26:11] [SPEAKER_02]: We have digital x-rays.

[00:26:12] [SPEAKER_02]: We want to be able to detect decay.

[00:26:14] [SPEAKER_02]: We want to be able to detect bone loss.

[00:26:17] [SPEAKER_02]: We also want to look at

[00:26:19] [SPEAKER_02]: pens and be able to understand lesions.

[00:26:22] [SPEAKER_02]: We also want to be looking at a standard of care

[00:26:24] [SPEAKER_02]: that's right around the corner

[00:26:26] [SPEAKER_02]: for every dental office.

[00:26:28] [SPEAKER_02]: In my practice, we're all specialists.

[00:26:30] [SPEAKER_02]: It's a group of specialists working together

[00:26:33] [SPEAKER_02]: to raise the bar

[00:26:34] [SPEAKER_02]: to keep the highest bar possible

[00:26:36] [SPEAKER_02]: for quality of care.

[00:26:38] [SPEAKER_02]: When does it go to the cone beams and the CBCTs?

[00:26:42] [SPEAKER_02]: So map out a future for us

[00:26:44] [SPEAKER_02]: for the way you're thinking about

[00:26:46] [SPEAKER_02]: these issues in dentistry

[00:26:48] [SPEAKER_02]: and the quality of care

[00:26:50] [SPEAKER_02]: and where you can take your company.

[00:26:51] [SPEAKER_01]: I think at least the development

[00:26:54] [SPEAKER_01]: of AI on images is a solved problem.

[00:26:59] [SPEAKER_01]: If it hasn't hit the market,

[00:27:00] [SPEAKER_01]: it's going to hit someone.

[00:27:02] [SPEAKER_01]: So I don't think it's even multiple years.

[00:27:04] [SPEAKER_01]: It's this year or next.

[00:27:05] [SPEAKER_01]: Right?

[00:27:05] [SPEAKER_01]: So I don't think that's where

[00:27:07] [SPEAKER_01]: the long term will follow

[00:27:09] [SPEAKER_01]: because it is something that

[00:27:10] [SPEAKER_01]: the models have gone good enough

[00:27:12] [SPEAKER_01]: people knowing how to annotate data,

[00:27:15] [SPEAKER_01]: how to get the outputs.

[00:27:16] [SPEAKER_01]: That whole thing has been streamlined enough

[00:27:18] [SPEAKER_01]: that for example, for us,

[00:27:19] [SPEAKER_01]: we can get a new model out

[00:27:21] [SPEAKER_01]: in a week or so,

[00:27:22] [SPEAKER_01]: get it through.

[00:27:22] [SPEAKER_01]: Then it is the checks and the balances

[00:27:24] [SPEAKER_01]: and regression testing

[00:27:26] [SPEAKER_01]: and all the things that need to happen,

[00:27:27] [SPEAKER_01]: putting those in place.

[00:27:28] [SPEAKER_01]: But it is pretty now straightforward

[00:27:30] [SPEAKER_01]: because we've built models with

[00:27:32] [SPEAKER_01]: or more foundational models as well

[00:27:33] [SPEAKER_01]: that can get through outputs very quickly.

[00:27:36] [SPEAKER_01]: Now in terms of what is going to happen next,

[00:27:39] [SPEAKER_01]: I think in dentistry,

[00:27:40] [SPEAKER_01]: we've just touched the surface

[00:27:43] [SPEAKER_01]: with being able to identify things on X-rays.

[00:27:47] [SPEAKER_01]: And our contribution to this industry

[00:27:49] [SPEAKER_01]: was quantification.

[00:27:50] [SPEAKER_01]: We introduced quantification into the market

[00:27:51] [SPEAKER_01]: and I think people followed afterwards.

[00:27:54] [SPEAKER_01]: But if you think about it,

[00:27:55] [SPEAKER_01]: that's pretty basic in terms of what we can do.

[00:27:58] [SPEAKER_01]: I think the next big unlock

[00:28:00] [SPEAKER_01]: is going to happen with treatment planning.

[00:28:03] [SPEAKER_01]: So how do you actually create treatment plans

[00:28:05] [SPEAKER_01]: based on what the user needs are?

[00:28:08] [SPEAKER_01]: How do you take in all the information

[00:28:09] [SPEAKER_01]: off the patient to come up with a better treatment plan

[00:28:11] [SPEAKER_01]: than any average dentists could come up with?

[00:28:14] [SPEAKER_01]: Maybe not a prosodontist,

[00:28:15] [SPEAKER_01]: but like how do you actually then get

[00:28:17] [SPEAKER_01]: to a prosodontist level as well?

[00:28:19] [SPEAKER_01]: But those are things that I think

[00:28:20] [SPEAKER_01]: will be the next big unlock.

[00:28:22] [SPEAKER_01]: And it's not easy.

[00:28:24] [SPEAKER_01]: There's a lot of research going on.

[00:28:25] [SPEAKER_01]: How do you actually do this well?

[00:28:27] [SPEAKER_01]: And then do it well,

[00:28:29] [SPEAKER_01]: taking into account the entire patient data,

[00:28:31] [SPEAKER_01]: not just a single snapshot in time as well.

[00:28:34] [SPEAKER_01]: And then the third is like,

[00:28:35] [SPEAKER_01]: how do we actually think about outcomes

[00:28:37] [SPEAKER_01]: and improving those outcomes as well?

[00:28:39] [SPEAKER_01]: We have started work around that aspect too

[00:28:42] [SPEAKER_01]: and we are working with lots of different agencies

[00:28:45] [SPEAKER_01]: to come up with the right ways

[00:28:46] [SPEAKER_01]: of measuring oral health outcomes

[00:28:48] [SPEAKER_01]: and then being able to track those.

[00:28:49] [SPEAKER_01]: So I think that's where I feel like

[00:28:52] [SPEAKER_01]: a lot of the next advancements will happen

[00:28:54] [SPEAKER_01]: to really help dentists provide the best care that they can.

[00:28:59] [SPEAKER_00]: Yeah, that's great.

[00:29:00] [SPEAKER_00]: Well, we set an intention at the beginning

[00:29:02] [SPEAKER_00]: of this conversation to learn.

[00:29:04] [SPEAKER_00]: And boy, have we been learning

[00:29:06] [SPEAKER_00]: in the last half an hour or so.

[00:29:08] [SPEAKER_00]: Order, we are at the end of our time,

[00:29:10] [SPEAKER_00]: but before we let you go, one last question.

[00:29:14] [SPEAKER_00]: How could our listeners

[00:29:17] [SPEAKER_00]: budding aspiring AI early adopters

[00:29:19] [SPEAKER_00]: stay in touch and keep up with what you're doing,

[00:29:22] [SPEAKER_00]: the research you're publishing.

[00:29:24] [SPEAKER_00]: What's the best way for them to get involved

[00:29:26] [SPEAKER_00]: in getting touched?

[00:29:27] [SPEAKER_01]: So we have our website overjet.com

[00:29:29] [SPEAKER_01]: or overjet.ai, both lead to the same website.

[00:29:33] [SPEAKER_01]: I think that's where we'll find more resources

[00:29:36] [SPEAKER_01]: as well as ways to connect with us.

[00:29:38] [SPEAKER_01]: Easy.

[00:29:39] [SPEAKER_00]: Well, thank you very grateful to have you today.

[00:29:42] [SPEAKER_00]: John, did any parting thoughts on this topic?

[00:29:44] [SPEAKER_02]: Very inspiring.

[00:29:47] [SPEAKER_02]: And as we look at the future, and what I think even your company

[00:29:51] [SPEAKER_02]: are addressing some real needs within the industry.

[00:29:54] [SPEAKER_02]: As you outline the future, once we get past this,

[00:29:57] [SPEAKER_02]: the quantitative impact that you're having on the industry

[00:29:59] [SPEAKER_02]: with reading x-rays, what you pointed out

[00:30:02] [SPEAKER_02]: about treatment planning, the subjective nature of it,

[00:30:05] [SPEAKER_02]: the complicated ability to put all of those synthesis

[00:30:09] [SPEAKER_02]: of the data together to come up with a plan

[00:30:12] [SPEAKER_02]: that integrates the different disciplines

[00:30:14] [SPEAKER_02]: and interdisciplinary dentistry is an area that's really

[00:30:19] [SPEAKER_02]: ripe for change and improvement.

[00:30:22] [SPEAKER_02]: So that would be quite amazing and very inspiring

[00:30:25] [SPEAKER_02]: to hear that you're going to get after that.

[00:30:28] [SPEAKER_02]: Very exciting.

[00:30:29] [SPEAKER_00]: Thank you.

[00:30:30] [SPEAKER_00]: Thank you for joining us, Huerta.

[00:30:31] [SPEAKER_00]: And we'd love to have you again with some updates.

[00:30:33] [SPEAKER_00]: Sounds like the pace at which you're

[00:30:35] [SPEAKER_00]: evolving the product and adoption is pretty amazing.

[00:30:39] [SPEAKER_00]: So soon you'll have updates to come back to share with us.

[00:30:42] [SPEAKER_00]: So extending an invitation to come back.

[00:30:45] [SPEAKER_00]: And talk to you soon.

[00:30:47] [SPEAKER_00]: Thank you.

[00:30:47] [SPEAKER_00]: Thank you.

[00:30:48] [SPEAKER_00]: Take care.

[00:30:54] [SPEAKER_02]: Thanks for listening to the Think Oral podcast.

[00:30:57] [SPEAKER_00]: For the show notes and resources from today's podcast,

[00:31:01] [SPEAKER_02]: visit us at www.outcomesrocket.health.thinkoral.

[00:31:08] [SPEAKER_00]: Or start a conversation with us on social media.

[00:31:11] [SPEAKER_00]: Until then, keep smiling and connecting care.