Your AI Injection

Measuring and Monitoring Plastic Surgery Results Using Crowdsourcing and AI

March 15, 2022 Season 1 Episode 16
Your AI Injection
Measuring and Monitoring Plastic Surgery Results Using Crowdsourcing and AI
Show Notes Transcript

This week we’re speaking with Dr. Daniel Gould and Dr. James Smartt. Dr. Gould’s a plastic surgeon and the author of hundreds of peer-reviewed journal articles. Dr. Smartt is an attending surgeon at Bucky Plastic Surgery and the CEO and co-founder of Love My Delta, a smart phone application that measures and tracks cosmetic surgery improvements over time. 

This episode discusses how AI is impacting the world of plastic surgery.  We dive into ways treatment efficacy is measured through crowdsourcing and AI and how both methods are essential to gather data.  We also discuss the app Love My Delta and gain insights into how the app is beneficial to physicians by helping improve patient outcomes and product efficacy. 

Automated Transcript

Deep: Hi there I'm Deep Dhillon. Welcome to your AI injection. The podcast where we discuss state-of-the-art techniques and artificial intelligence with a focus on how these capabilities are used to transform organizations, making them more efficient, impactful, and successful.

Welcome back to your AI injection. This week, we'll be digging into how AI is impacting the world of plastic surgery. We're speaking with Dr. Daniel Gold and Dr. James Smartt, Dr. Gold's plastic surgeon, the author of hundreds of peer reviewed articles. Dr. Smartt is an attending surgeon at Bucky plastic surgery and the CEO and co-founder of love my Delta, a smartphone application that measures and tracks cosmetic surgery improvements over time.

Uh, Dan and Jim start us off by telling us a little bit about how you guys know each other and how and why you got into plastics.

Dr. Gold: I'll just kind of jump in and say, Jim was a mentor to me at a very young part of my training. So I was actually a visiting medical student who had come from Baylor college of medicine in Houston to go to the epicenter of chronic facial surgery, which at the time was UW in Seattle. And Jim was the cranium facial fellow. And, you know, together, I basically. The sub intern, the person at the bottom of the service. And I was learning about craniofacial care of infants patients with cleft lip and palette. And Jim was the head of the service at the time, or essentially the top level of kind of the training echelon at UW. And we just kinda clicked on a level that was, I think, you know, a level of friendship, but also respect. And I was getting tons of mentorship from him just entering into plastic surgery. But then Jim and I went on rounds one morning. And Jim, you want to talk about what we were?

Dr. Smartt: Yeah, I mean so deep and this is real. I mean, it's as organic a connection as you can possibly make. So at the time Tomlin bay was an attending as well, and Deep was involved in the project that is sets. I had seen Tom on numerous. While I was there and I said to Dan, I said, you know, I'm involved with this crowdsourcing endeavor and I can't help, but think that we can apply these methods to plastic surgery.

Dr. Gold: I should say one really important concept. We were looking at children who had cleft lip and palette, and we were saying, how do you know what's a good result? You know, the concept of plastic surgery is that results matter. The way things look matters, the visual scale, the visual assessment of a cosmetic or reconstructive outcome is the key. So we were on rounds and I said, Hey, Jim, how do you know what's a good outcome? Like, there's all these different ways to do this surgery. You have some patients that look really good and some that look really bad. Is there some way to really come up with a common denominator or like a batting average to figure, go what the scores are of the outcome?

Deep: Tell us a little bit about what is done historically within the plastic surgery community to like say I did a good job. Like I eliminated the cleft palate and, and it looks great or it looks normal or it doesn't like, how has that historically been that?

Dr. Gold: Well, traditionally there are different ways to look at those outcomes. There are scales that are like visual scales that look at anatomical relationships and equations that identify. The restoration of the natural form and function, and those basically subset or differentiates a static areas of the face. And for a lip, for instance, you could be looking at the white role or the Vermilion or the distance of the philanthropic column. All these are anatomic names for things that we see and plastic surgeons, traditionally, look at those things and say, that's a good outcome or bad outcome. And in the past there were articles. Where they looked at it and said, look, these are my results. I did 200 of these, and this is what I think it looks like. And then they had other plastic surgeons look at it and they said, great job. But that's not really assessing what the general population thinks, what the crowd thinks. And there's no built-in tools to further hone those results. And to really understand the visual knowledge that we see in the before and afters of these patients.

Dr. Smartt: Yeah, we raised our, you know, friends and family round. And that started in may of 2020. That was the time when kind of love my Delta came together. It's important to put in context of why our crowds and why are machines a good fit in my mind. And it's because, you know, we want to produce visual knowledge and visual knowledge is scarce. It's scarce in plastic surgery, it's scarce in aesthetic medicine. And if you're going to build a business. Uh, I think it's a good idea to be able to create something that's scarce. It puts you at a competitive advantage, right? So that was the goal.

Deep: So walk us through, what exactly does it mean to like take measurements and sort of in an objective way, assess the efficacy of some kind of.

Dr. Smartt: Uh, you know, this is where companies and physicians and drug and medical devices are asked to perform and when they do so, the, the, the rigor comes from being able to perform measurement. And Botox is a great example. For instance, you know, when that idea first started, it seemed outlandish. You know, you're going to inject a toxin in the humans, and you're going to try to track a visible difference. And how their wrinkles look. Right. And in order to do that, you needed to perform measurement. You needed to have reliable measures of what does it look like when you have hardly any wrinkles? What does it look like when you have a whole lot and somewhere on that spectrum, you had to be able to essentially for the guys in AI, it's a sorting task, right. And you have to sort. These things into buckets immediately. When you look at the, the measurement tools that are used by, uh, companies that are going through the regulatory process, they use these things called visual scales, which to a statistician. Is what's called a latent variable model. And that's when you know, you look at a person, you look at how bad their wrinkling is. You put it up on a scale. That's been made by a group of statisticians, and then you come down and you arrive at a. Right. And this is what we do when we're humans. And on the AI side, if you want to perform the same type of measurement, these are classification tasks, generally in image analysis. And so the two things, generally they are essentially parallel means to get to this idea, which is measurement, right. We want to make that our brand very simply.

Deep: So going back down to something that you said you were talking a little bit about the sort of classical approach and I, and you were talking about how it was mostly plastic surgeons themselves, correct rating codes, right.

Dr. Gold: You know, there's so many written articles where it's plastic surgeons evaluating other plastic surgeons outcomes. And if I have to be quite Frank with you, I find that, you know, kind of pedantic, like it doesn't make sense. There's one article where. I put his outcomes out there. And how does fellows grade the results it's like clearly biased, right? Correct. And this is a key concept there's bias in everyone. And if you can average out those bias over large points of reviewers, you can control for the bias that exists in our society. That's kind of a fun research side of what we do. And that's part of the secret sauce of cultivating the right crowd. And I think that's the importance of AI because AI is the interface. Where we identify those biases. We identify trends and we either add to them and then amplify their effect, or we cut them back and decrease their effect over time. And that's critical because that's how you sort the noise from the real data. Now, the question that you're asking is why is it that we don't have a better system? And the answer is because we haven't had scale nowadays we have scale and Turk, we have all these other different platforms that exist to be able to really explore these kinds.

Deep: But even in, in the seventies, you know, we had cameras, we had humans, we could have went around with cameras, shut photographs, taken measurements, and asked regular people.

Dr. Gold: It could, you know, right. That's what they would do. So there was a study where there was a burn unit in USC and they wanted to look at outcomes and identify the best place to take a skin graph from. So they wanted to ask people, if you had to have a part of your body where you take skin from, where would it be? So the fellows went to UCLA. Took photos of where it would be or would be areas and surveyed the general population. It's where you want your scene graph harvested from most of the people that are talking to our young girls who like want to wear skirts and things. So they're saying behind the posterior thigh or lateral thigh, but there's bias in who they're asking what they're asking when they're asking. And that's like not a scalable experience because maybe there's a hundred people. They asked that day and halfway there's probably cultural differences to cultural differences, differences in background, in, in economics, in the physical space that people live in that dictates and determines their daily activities. And those are the exciting parts of what we see on the other end.

Deep: Jim, take it back to LMD now. So we've got this ability to identify a part of the body that we want to get objective measurements from what exactly has to happen to get those objective measurements. And what's the role of the machine. And.

Dr. Smartt: That's interesting. And it speaks very much, I think, to Dan's last point, which is that, why have we not been able to do this? Right. We had cameras in the seventies and it mostly has to do with costs and it has to do with the burden of analysis because the burden of analysis. For researchers in the past has been enormous. Like the study that Dan was just talking about, you know, physically going out there, co-leading all the results, getting the research team together. And this is exactly the same burden that we see that pharma has right now in terms of going out and performing studies like this. You see what they're spending in the later phases of these trials. So hundreds of millions of dollars. And I think, you know, the promise of AI and, and crowdsourcing together is that they're both mechanisms to perform measurement at scale. Right? The big longterm goal of Love My Delta is to supply the check that the throughput human check on what are the AI's that will surround us in the future in many respects. Okay. So if you have an AI right now, the burden of validating, it really falls on the humans now. And how are you going to do that again? Well, it's very much to solve the same measurement problem that Dan was alluding to with the burn study, but it's essentially to do it in a modern format. That's what we want. The system that is able to accommodate exactly that kind of throughput and never really strays away from its origin in the sense that it's always a combination of human inference and machine learning.

Deep: Tell us, why does a patient care that LMD exists?

Dr. Smartt: Well, I think that patients organically care, I would argue that the birth of companies like RealSelf are examples of this in the sense that there are knowledge disparate. I think that the origin of real self was an incredible idea because it was trying to solve a knowledge disparity, right? If you're a patient and you want to know, is this surgeon, you know, this surgeon is going to cut me. This surgeon is going to do something to my body, and I want to make sure this is not like buying a car. When we first saw these knowledge disparities, it was in the context of just gathering information in its most basic sense. Right. And that's where the mobilized communities really made sense. And they still do. There's a more rigorous version of that and it's called patient reported outcomes. And this is something that is becoming a real trend within plastic surgery and it's incredibly valuable knowledge. Right. But the thing that it misses is the connection between. How people feel about how they do and ultimately what happens to their appearance in the change. Right? I would take it further. I would say love my Delta is more focused on transparency in that moment because all these other systems that have come about before now have been systems that have been designed for marketing tools. And at this point, we're looking at outcomes. We're trying to learn from those outcomes, become a way to evaluate outcomes within a system like an FDA approved process, and then to translate those outcomes to patients.

Deep: So two-part question. What is, do you think that there are significant. Visual outcome differences across physicians. What are some of the causes of that? But as a physician, when the output of your work is assessed, is that a threatening thing?

Dr. Smartt: So that's a multi-part question. That's pretty complicated. But what I would ultimately say is that I think this is a humbling process, that there is a demonstrable difference that is attained by most surgeons that it in general. Yes. A reasonable outcome that is attained by humans and that a lot of us communicate together and we generally get something that's kind of similar. There is a true reversion to the mean, right? However, the, you know, the thing that makes. AI and crowdsourcing relevant here is again the burden, right? So if you were, even if you were a surgeon, let's say 20 years ago, and you wanted to compile your own results, and even if you could get all the photographs together, there would still be an incredible burden upon you to develop. A standardized metric or scoring system upon which it would be evaluated and then to apply it with some throughput. And I'll just give you an example, you know, this has been done in plastic surgery and it took incredible, um, energy and organization on the part of the people who did it. So they would get for instance, panels together, right. That you would have at academic conferences. And maybe you would literally get paper out and you would have examples of pre and postoperative states and they would get rated, but it hardly ever happened. And it was an incredible burden when it did. And so what we have now is two mechanisms to come some burden. The burden is a burden in time. It's costly. It's inefficient generally. And just to give you an example, like lots of these things happen when people had to get physically co-located and rate things, everybody-

Deep: And wheen you look at that result. Do you see a big difference across positions for a particular procedure?

Dr. Smartt: I think that that question, honestly, Totally remains to be answered. Um, I can tell you some of our early work in love, my Delta, some of it's not yet published, but I can tell you what it looks like. And if you take a group of facelift before and after results that are on the internet that are publicly available. And you interrogate them with a scale that is validated. What you find is that there is a grouping of results that vary in their outcome from very, very rejuvenating to not so rejuvinating, but the vast majority tend to sit in the middle and the overall magnitude of that change. Is not always profound. It's something that, that, again, there is kind of a reversion to the mean

Deep: That's kind of different though, right? So if you look at Hollywood actress X, that everyone sort of agrees had a botch thing, right. That could be that the plastic surgeon did a terrible job, but it also could be that that was destined to be a terrible outcome. And that if you swap them in with 10 other plastic surgeons, that they would have all wound up with a similarly poor outcomes. So how do you tell the difference?

Dr. Smartt: Yeah. It's a concept here is that there are people that do face ups that add volume. And there's a study that looked at younger versus more traffic main to add volume, big fat from their body, add it to their cheeks or to parts of their face.

Deep: Oh, because they were too golf looking or something.

Dr. Smartt: They lost younger. It makes you look younger because you look like the baby, you know, more full face that doesn't necessarily mean that you look better. That doesn't necessarily mean that you look more attractive. So there are studies that have looked at attractiveness versus age, and those are things that we can parse out with what we're doing. We can say to the crowd, does this person look more attractive to them? Look younger is the volume. We can do visual scales. We can narrow it down to the periorbital aging, the mid-face to the cheek, you know, to the cheek, into the lower mandible and jaw. And I think that if you subset those, then you get real knowledge from the photographs that you're seeing. There are so many questions that have not even been addressed because the burden was so high before. And that's the promise of both of these areas

Deep: You're listening to your AI injection brought to you by xyonix.com. That's x-y-o-n-i-x.com. Check out our website for more content, or if you need help injecting AI into your organization.

So one thing is we've been talking about kind of crowds and AI as if they're synonymous, but they're really not like the, the humans are helpful in providing training data for the machines. Humans have an inevitable latency, meaning that it takes them a while to actually assess something, whereas the machine can do it in a few milliseconds. Right. So what's the goal of the machine learning system or the AI system here.

Dr. Smartt: There are two systems that perform measurement at a scaled and both of which have different characteristics and different, different pros and different cons. Yeah.

Deep: This is a very unique setting for our audience. Like almost always, you want the AI system to drive as much as it can with effect. And the humans are really there to just be training the machine systems and abstinence. But in this context, you're saying something different. You're saying something like, actually I want. Input of a specific crowd of people. And then you're saying, but at the same time, I might also separately want the input from the model.

Dr. Smartt: It's one of these things that there is a fundamental human appreciation to beauty and to aesthetic results. And there's also a human value to that, right. That we can very much put in dollars because people spend and, you know, uh, beauty. Uh, self care medical aesthetics, plastic surgery, that entire category is going to eclipse 1 trillion in global spin very shortly. So it is something that's a great value to us, but our subjectivity in assessing that value and that will never go. And so to my mind, it's the ultimate context in which to test for bias within AI, because does it actually reflect the preferences of the people that are valuing it. And I think this is a beautiful co-mingling of two systems because AI it's incredibly valuable. It has great throughput when it is functioning with known norms that drive its analysis. It's actually doing something that's totally human. And a great example is facial symmetry. Right? So things that are symmetric are well-known to be more attractive. They're probably also incredibly easy to differentiate. Uh, using an AI things that are symmetrical or not, right. That is very human, but other aspects of appearance are much more culturally defined, right? There are a litany of character traits that have been found to have. Different appreciation of across different cultural groups. So there's an example where, well, if you're going to build an AI and it's going to, it's going to actually be able to make choices or sort things into buckets that are then reflective of the truth that you would garner from a particular group of people will. Maybe that has to be informed also by. Humans very much acting in a context that looks like labeling. Right. And it's something where I think, you know, the brand you want to build is one that they operate in parallel.

Deep: It's an interesting question because I think it depends, but if I have one human. And one outcome. And I put that to a cost-effective number of humans. So like maybe I can afford to put it in front of 50 humans. And let's say, you know, there's some cultural representation going on. Let's say it's America. And I want a cross section of America. So I put that in front of 50 people, but meanwhile, I could have a machine learning system that's been trained off of a billion people's input. And it could have a way broader representation and it could have learned. And ultimately, you know, outperform that small subset of 50 actual humans. I would argue that if we do this right, like the, the machine is going to end up being a better representative sample than the 50 that you pay.

Dr. Smartt: I completely agree in the long run view that on many of these topics, ultimately there is an AI product that could be the optimal endpoint. However, when you look at what we're dealing with nowadays, which is basically the first iteration of computer vision models that are populating spaces like beauty and whatnot, I think you really have to take pause and ask yourself at this iteration in our development of MLS. Is it something that is absolutely full-proof or do we have to have some mechanism that has high enough throughput that we can act on and really wouldn't that be a valuable brand, moving into the future within beauty and aesthetic medicine to say, you know what we represent this, and this is something that we, we intend to commingle. And there are probably settings where yes, the AI absolutely wins on every metric. But there's also, I just don't think that that is going to be ultimately compelling. There's going to be this. I don't want to use this uncanny valley phrase, but there'll be this moment where you're going to say, well, what does that all mean to how I feel are more sensitive?

Deep: I have to what people tell them then to what machines tell them, right?

Dr. Smartt: And this is a verdant, this is verdant soil. This is exactly what you're saying. What's the intersection between artificial intelligence and machine learning and crowdsourcing. And I think that what, what you're alluding to the fact that those cultural differences within groups within Glades are very different than what you see in the machine learning iteration of billions, of people or hundreds of millions of people. And I think that's where all of the great research is going to be done. And that's where we're going to learn them. About cultural appreciations and norms for aesthetics about princes in markets differences in demographics. This is the value of LMD. If we can compare those two things and say, you know, you're a consumer in this market and south Florida and the United States and an area of this area, this is what beauty is in this region.

Deep: Yeah. And you could, you can imagine multiple models and multiple vantages, like trained. I want to switch directions a little bit, but in the sort of FDA world, you know, you have a treatment, you go through your clinical trials, the FDA sort of stamps it and says, Hey, this thing's all good. And then the treatment goes out into the wild and the creator of the treatment now is responsible for assessing. Sort of evolution in the wild, the side effects that emerged in the wild and reporting those back to the FDA. And one of the things that I found really fascinating about LMD when I sort of looked at it early on, was this idea that achievement like let's take Botox, for example, can be approved. But there's all these potential variations in terms of the particular batch that might get created or a particular institution and how they apply it. And you might, you might have like problems happening out in the field. So talk to us a little bit about like, what is the potential for taking a post-approval treatment and then tracking it in the wild and like, what do you get from a medical advantage that you don't get with the self reporting system that we have.

Dr. Smartt: This is huge. I think this is one of the most valuable examples that we have at love my Delta. So exactly what you're saying within the spectrum. Let's talk about neuromodulators. There are different neuromodulators that are made by different companies. They're diluted in different ways. They're effective at different dilutes. Differentially they're injected in different places and they can be injected for similar effects in different anatomic regions in the face. So if you're a large consumer of neuromodulator, let's say a large medical spa chain and you have 30 locations and you want to identify whether or not there are variations in quality of outcomes or ways to improve the throughput of your treatments. And the patient has. Or ways to improve your bottom line by, by using the right amount of product in the right place at the right time, then Love My Delta can empower you because it can allow patients to interact in a way with their visual knowledge, by taking photos, looking at the areas of their face that are changing the treatments and report those back to the provider or in a large scale back to the system. And the system can look at it and say, We're doing a really good job of treating the glabella and the forehead in our practices, but our crow's feet areas are not really as well developed. And why is that? Is it because we have a deficit in injector education? Is it because we're using the product and it from a different area in that area, in that region,

Deep: You might have a bad batch or something.

Dr. Smartt: Well, matches are key. So I'm glad you brought that up because there's not a lot of understanding about what goes on in batches, in the boat, up in Botox. Or modulators. And I think all surgeons and all med spots, I've seen a patient that comes in and they say, I don't think it was enough and there's a totally animated. And maybe it was that the bachelors left out or that it was like you said, a bad scape, bad batch. He came to the company. And to me it all falls under this giant category of risk management. Right. Which fundamentally something. I think that, you know, there's various realms where AI versus crowds that are going to be the optimal mix. Right? So when it comes to marketing and human appreciation and beauty, maybe the optimal mix is more crowds than AI. But let's take low frequency events that are catastrophic right now. And that's where risk management is critical. Maybe that's this case where, you know, and a good example with neuromodulators are injectables is something that's called a vascular occlusion. Okay. And this is when you inject a filler into someone's face and there's an area that. It, it mistakenly goes into an artery and then the area that, that artery perfuses is occluded and there's not enough blood supply. Right. So it's a low frequency event. It's really uncommon and it can be managed very effectively, but the clock is ticking as soon as it happens.

Deep: So you want to get that patient into a physician as soon as possible

Dr. Smartt: Exactly. And it's very, very treatable. Most of them are very treatable, but, but like I said, there's a time component, right? Like, you know, in this setting, here's an optimal place where an AI would be perfect. Right? You get a computer vision model that is trained on a series of patients. Who've experienced an occlusion, and then you apply that to mobile technology and put it on a phone.

Deep: Yeah. We didn't actually describe quite the context of the use to that. You know, the patient's got a, an iPhone they've sort of guided through. They're able to like take a photograph, um, particular imagery likes a forehead lines or the crow's feet or something. And in that context, that data goes up. Machine learning models are applied. Crowds might be applied. Um, but we could in theory, detect this particular case that you're describing. Right. And get that patient-

Dr. Smartt: and a great example where an AI is really the only solution that works because even a crowd, study's not gonna, you know, there's not even if you had in theory, this like incredibly engaged crowd that was available up to the second. It's like a, you know, the Turk of aesthetic medicine. Right. And we'd kind of love to create that, but that's probably not actually available right now. And even the signs of that are something that you're not quite sure whether humans would pick it up. It's a perfect case, right? Where an AI that sits on the phone would be able to detect this immediately. And these are the kinds of things we're seeing, you know, In fields like ophthalmology and dermatology that already exists. Right?

Deep: I feel like this template is really important and it transcends plastic surgery, which is. You have someone who may have some kind of issue. And there's inevitably a long time window between when they're going to see a healthcare provider, but if you've got a device that they're interacting with daily or regularly, and now you've got the ability to detect that new suspect lesion that's on their hands.

Deep: Or, you know, in this case, you know, with a toxin that's sort of causing a strange muscular reaction or something, that's a really powerful-

Dr. Smartt: absolutely. That really speaks not just to AI or plastic surgery. That's really mobile. Right? There are so many cases of adverse events in plastic surgery. Where a crowd or an AI could discern that there is a change that is worrisome, but when you combine both of those modalities and then you put mobile into the system, there you have something that, because machine learning or crowdsourcing, if the measured outcome is not seen. Neither one, acts on it anyway. Right. So, and that's the beauty of incorporating mobile technology in the LMD is that, you know, our patients spend 0.001% of their lives in our offices. Right?

Deep: Yeah. And I think this is a pattern that we're seeing in a lot of healthcare startups. It's a huge part of the physician. Patient interaction is built off. Very fundamental assumption that patient comes in and spends 15 minutes with the physician and then disappears for a good chunk of time. And it turns out that there's like this whole dark space that healthcare providers are operating within, where they don't know what happened to the patient during that time. And now we have the ability, you know, whether it's, you know, via a phone or it could be Alexa or Google home, it could be like security cameras, but we've got the potential. To have the, you know, the care providers ability to like draw and gather input from all that dark time that they currently just don't get

Dr. Smartt: Well, let's, let's call it the arc of engagement. Let's say it's it's, you know, from the minute, like you said, patient has high blood pressure and they go to see their doctor. Doctor prescribes the medication and says, see me in three months, patient goes to the pharmacy patient fills the medications, misses the medication once or twice, she's a nurse doesn't nurse says, oh, we can refill your meds, goes to a Walgreens, checks their blood pressure. Three months later, they've had all these things that have happened that we weren't a part of. And that's the beauty of, Love My Delta is that those patients let's say, make it a patient. That's doing neuromodulator now they get their injection two weeks out. They have residual lines. They're unhappy. They haven't said anything to the provider. They didn't go back into doing anything. Three months later, they've dropped off the books. They say, Botox doesn't work for me. And it's like, no, you just were undertreated. Or maybe they got a great treatment. Or maybe you had a complication have ptosis in their eye, but they never said anything. You're right. You've got a whole arc of engagement. And every one of those moments along that arc is an opportunity to improve the quality. To gain knowledge about what we're doing and to make it better and to translate that into visual knowledge and to a consumer system that really would benefit from the outcome.

Deep: And that to me is just a huge part of the future of, of medicine. Is this ability to go into what you're calling the arc of engagement and extend that, because I think there's just so much potentially can go wrong with this 15 minutes, a month interaction style that physicians have with their.

Dr. Smartt: I want to dispel one thing. So, and I think we've given aesthetic medicine and plastic surgery, a little bit of a rough ride thus far. And if you're a plastic surgeon, you are very engaged with your patients. Okay. It is, it is most likely to realm of medicine where we have the most responsive offices and the most responsive manner in the way that we deal with patient satisfaction. Right. It is our currency. If patients are happy, we do well. If they're not, we it's challenging. So we may not be great at assessing our visual outcomes, but we are really good at knowing what makes people happy. And this is why this engagement thing I think, and this, you know, we're creating this brand, this, this very subtle combination of mobile. AI and crowds and there's going to be a mix that's right. For every setting. But the one thing I do not think is going to be a problem is more engagement is not going to be a problem in this setting because we already have. A ton of engagement. I just think that this is something that will enhance a relationship that's already generally pretty good yet. In other fields, there are real gaps, right? I mean, when you think about the things that we could also deploy this system toward. There, there are tons about cones that have a visual component that could, could be measured better and through their life cycle. And I think that, you know, it's just, there's so many exciting applications for this, that it really it's mind blowing on a level.

Deep: I'm going to ask you guys another question. You know, there's a lot of potentially negative connotations of plastic surgery, but I want you guys to speak to like. What are the health reasons that you see people getting plastic surgeries and like what percentage of those, it makes up kind of an overall average plastic students care.

Dr. Smartt: So, well, I mean, this, this is a fundamental distinction in our discipline, right? Is it cosmetic or is a reconstructed. You know, this is where the great institutes of plastic surgery originated, whether it be places in Britain, American centers like NYU. And it was in these places where you had the reconstructive challenge of maimed people. And so that is fundamental to plastic surgery. And then, you know, that evolves into in better times the burden of cancer mostly. And so you're going to remove parts of people's bodies and you've got to figure out some way to cover defects. And you've got to reconstruct things nowadays in ways that aren't sometimes just good, but even better than how you start. Plastic surgery is very much had this very advanced set of tools and always has. And this is for a lot of us, one of the kind of blessings of the job is that it's interesting from both a reconstructive and a cosmetic standpoint. And so that tension always follows us. Both Dan and I have done these things in our career. And so there's this weird line that always gets towed between, you know, beautification and just sustaining.

Deep: I'm just curious from a, from a medicine vantage, like is the line. Anyone can get anything anytime they want, or is the line there's genuine, maybe psychological burdens being carried here, therefore. Or there's physiological burdens. And how do you parse that as a physician? And then we can get back to AI.

Dr. Gold: Those are concepts that we've explored. Like in my research, I've looked at reconstructive and cosmetic things and I've used, I've used crowdsourcing to help us identify appropriate outcomes in breast reconstruction, in erectile dysfunction, new treatments for that as for hand reconstructive surgery, pediatric reconstructive. These are the things that we've all explored. And I think that we need to blur the lines a little bit here because you know, what you're really asking is are there procedures that are cosmetic that are beneficial? And I think that if you look at mommy makeover, abdominal classy, we're restoring abdominal competence. We're putting muscles back together. We're doing a pubic lift, which has been demonstrated in four studies to improve your inner outflow incontinence and one study to improve your sexual function. There are functional benefits to what we're doing. And I think that those are important to understand because that's why. Cosmetic surgery or whatever is important. And that's why we need a space where we can evaluate.

Dr. Smartt: Now that we have these tools available to us, whether it's reconstructive surgery or whether it's cosmetic surgery, the tale to be told is how do you, how do you perform measurement at scale and value that visual change? Right? So really, and deep, when you think about recon versus cosmetic, there are just as many use cases. In the world of reconstructive surgery as there are. I mean, I can, I can list you the litany of companies that I've already been talking to, which are essentially drug or device manufacturers who create products for the reconstructive surgery market and have the exact same challenges that we talked about for the beauty industry. So, you know, that's, that's where the platform, it can Excel across all those domains.

Deep: This has been an awesome conversation. I'm going to ask you both to take a crack at this final question, which is in 10 years, what areas of plastic surgery are altered that, you know, by AI on that kind of a time arc our patients better off. And if so, how?

Dr. Gold: Yeah, I think that's a great question. So I think that in 10 years, these different platforms will allow surgeons to evaluate their own outcomes, objectives, to improve and to continue to improve the quality of care and the quality of the results. And the patients will see it from a transparent perspective. They'll know, oh, these are, this is somebody who's at a 70%, but they're improving. This is somebody who cares a lot about their results because they've moved. They've used the software. To improve from a novice to an expert in this technique, or there'll be somebody who's been verified or validated by several systems and maybe in reconstructive surgery, there are reimbursements tied to the ability that you have to demonstrate your improvements in these systems. And maybe overall, the entire population has a better understanding because of that transparency of the importance of what plastic surgery means to a community and the importance of the outcomes and how they're judged. And I think. Outside of plastic surgery. There's a whole cosmetic awareness that will be different. It'll be everything. It'll be people who do brows, fingernails, you know, haircuts. It's going to be a whole different realm of beauty. That's utilizing the technology that we create in the first wave in order to identify. Aesthetic outcomes and norms, and to really improve the quality of care in those other industries. So I think there's going to be a huge opportunity for growth. There's going to be a sprawl. We have to show that this technology is effective and useful in this medical capacity, but this is like the gold standard. And if we can do that, then I think there's a huge opportunity ahead for real visual knowledge, to be assessed about aesthetic outcomes to improve them. That's the whole point.

Dr. Smartt: Measurement. And that scale is I think the next phase. And if we can get this kind of secret sauce of delivering value for people and also providing measurement that is, uh, you know, sometimes objective and maybe not that fun, but it's definitely value. Right. And so I think that, that as I see it, it's this we've kind of, we've gotten through the appreciation of opinion, subjectivity and communication. That was the first phase. And now we're moving into measurement and I love this project. It's been a total blast to work with you all. And just, if we can combine that duality of the throughput you get with crowds and the throughput you get with AI, I think we're going to find a lot of solutions to measurement problems in the next 10 years.

Deep: That's all for your AI injection as always. Thanks so much for tuning in. If you're interested in learning more about the intersection of AI and healthcare, check out some of our recent projects on our website is xyonix.com/industries/health that's xyonix.com/industries/health. If you enjoyed this episode, please feel free to tell your friends about us.

Give us a review and check out our past episodes at podcast.xyonix.com. Or on your regular podcast platform, that's all for this episode. I'm Deep Dhillon, your host saying check back soon for your next AI injection. In the meantime, if you need help injecting AI into your business, reach out to us at dot com.

That's xyonix.com. Whether it's text, audio, video, or other businesses. We help all kinds of organizations like yours automatically find an operationalized transformative insights.