Пълният препис на #10 - Skill Erosion in the Age of Medical AI

them or box them off and show them to you. 00:02:53.788 --> 00:02:58.891 And then let's take the tool away and see what happens. 00:02:58.979 --> 00:03:00.447 After some period of use with the tool. 00:03:00.961 --> 00:03:03.689 Yeah, they got to use a tool for like three months, gotcha okay. 00:03:03.689 --> 00:03:07.407 So now, three months goes by, they stop using the tool. 00:03:07.407 --> 00:03:09.252 Yeah, they got to use a tool for like three months, I gotcha okay. 00:03:09.252 --> 00:03:17.957 So now, three months goes by, they stop using the tool and they see okay, are these trained clinicians, these endoscopists, these surgeons, able to still detect these precancerous lesions? 00:03:17.957 --> 00:03:22.944 Because the whole point is you want to remove them, right, you want to find them and you want to remove them to prevent cancer. 00:03:26.008 --> 00:03:39.697 And drum roll please what they found was that the adenoma detection rate decreased by like 6%. 00:03:39.697 --> 00:03:39.798 Wow. 00:03:39.798 --> 00:03:56.509 So by that metric, it seemed like the doctors at the end of the study without the AI, after having used the AI's assistance throughout, when they did not use it, at the end, their job performance in, specifically in the adenoma detection rate metric went down. 00:03:57.111 --> 00:04:00.544 Yeah, exactly, and so that's this concern around. 00:04:00.544 --> 00:04:03.091 Did they de-scale? 00:04:03.091 --> 00:04:05.021 Yeah around, did they? 00:04:05.021 --> 00:04:06.384 Did they de-skill? 00:04:06.384 --> 00:04:26.495 And I guess, kind of in parallel with that in my head, is this question of like, how bad is it to depend on AI tools like this, because other papers I've looked at have actually shown that AI helps you find more adenomatous polyps. 00:04:27.120 --> 00:04:37.901 Yeah, and I think that was the big motivation for this paper was that prior work, both in endoscopy but also in other applications, found that AI assistance can actually improve the quality. 00:04:37.901 --> 00:04:49.105 And we, you know, we talked about that with diabetic retinopathy in our first episode, right, when we talked about the AI being really good at finding certain things from images and being really able to help out. 00:04:49.105 --> 00:04:56.625 So the fact that it was able to do some assistance and actually improve that might have suggested that maybe it's a good thing. 00:04:56.625 --> 00:05:07.072 But what we're seeing now from the study is that, yes, maybe that's a good thing, but it seems like the doctors have lost their own ability to do that effectively, right? 00:05:07.072 --> 00:05:10.411 That's kind of what this study is suggesting. 00:05:10.980 --> 00:05:12.305 Yeah, and it's an observational study. 00:05:12.305 --> 00:05:27.678 It's not a huge study, but that's like the question that it's raising is you know, in just three months did these providers lose some of their skills and then so it doesn't matter, right? 00:05:27.819 -->

00:05:27.939 right. 00:05:27.939 --> 00:05:33.620 So let's take a pause here for a second and think about what that provider skill might look like. 00:05:33.620 --> 00:05:46.579 Like, can you walk us through what it might be like to do one of these I don't know colonoscopies, like what is the doctor doing and and you know, and then we can talk about the AI use in connection with that. 00:05:46.620 --> 00:05:55.925 Yeah, so you know they've got a scope in their hands, they're putting it in, they're taking a look at your colon and they're visualizing the inside there, right. 00:05:55.925 --> 00:06:03.226 So they're they might have to suction, they might have to press on areas they where the colon curves. 00:06:03.247 --> 00:06:32.189 They're going to have to, like, kind of turn around and look backwards, um, and they're trying to see if there are any abnormal growths there and this is all through like an endoscopy, like a little camera light thing, exactly yeah camera and light they put in, yeah, and they're observing it on like a screen or something, exactly right right and they're controlling that're controlling that scope and deciding where to look, how to look, what else they can do along with it to get a better look at something right. 00:06:32.511 --> 00:06:35.617 Right the. 00:06:35.617 --> 00:06:48.550 Ai is not doing that piece of things looking at what was actually displayed on the screen and flagging whether or not it might be an adenomatous polyp. 00:06:48.870 --> 00:06:49.071 Right. 00:06:49.071 --> 00:07:02.074 Right, and just to be clear, we're not sure exactly how the AI was used in this particular observational study, and I think that's something that I want to get into at some point as well, but this is good to understand from a contextual standpoint. 00:07:02.074 --> 00:07:15.505 As to the task here is not just looking at an image, but it's a very active task, um, unlike other imaging type tasks where you take one picture and you can then let the ai do whatever it needs to do. 00:07:15.505 --> 00:07:18.077 Here taking that one, there's no one picture. 00:07:18.218 --> 00:07:38.855 It's sort of a exploration, in some sense, and reasoning as you explore, deciding where to go next and what this means and what this could mean, and then doing some tests, pushing on it and stuff, and so that's yeah, like I want to pause here, or I need to suction because I don't have a good enough view, right, or I like I want to turn, or turn back and take a look at this from a different angle. 00:07:38.855 --> 00:07:47.158 So your endoscopist is is actually actively involved in that whole process, right, and it's active video that's happening. 00:07:47.158 --> 00:07:51.649 You're right, it's not a single image that can be sent for detection. 00:07:52.115 --> 00:07:54.560 Right, right, right, right, yeah. 00:07:54.560 --> 00:07:55.884 So that's very interesting. 00:07:55.884 --> 00:08:01.745 And of course, our metric here is the ADR, or the adenoma detection rate, which is sort of an end metric. 00:08:01.745 --> 00:08:02.877 Right, it's at the end of it all. 00:08:02.877 --> 00:08:13.762 It doesn't account for all the specific actions that the doctor is taking, so we don't know what skills specifically are being de-skilled, so to speak. 00:08:13.762 --> 00:08:25.035 Is it that the doctor is not looking enough because they're trusting what the AI is saying, or is the doctor not checking what the AI is saying and looking enough but not checking what the AI is saying, and looking enough but not checking what the AI is saying? 00:08:25.035 --> 00:08:30.187 It's unclear, I think, from the study, what specifically the modality was of that AI use. 00:08:30.187 --> 00:08:35.366 But if it's okay with you, I'd like to jump in a little bit and talk a little bit about that, a little bit more. 00:08:35.634 --> 00:08:39.461 About human-AI interaction, because I think you're right, that's what this problem is. 00:08:39.461 --> 00:08:42.885 It's not just an AI issue. 00:08:43.455 --> 00:08:45.423 Right, right, right right, and we have the issue. 00:08:45.423 --> 00:08:52.779 You know, we have the sort of the overarching issue of a so what, the doctor is getting descaled, and maybe we'll return to that towards the end of the episode. 00:08:52.779 --> 00:09:01.284 But I think the the question that I have in this is that what exactly is the doctor and the ai doing together? 00:09:01.284 --> 00:09:07.003 And there's a whole field of study human AI interaction, human robot interaction. 00:09:07.003 --> 00:09:22.644 There are thousands of researchers, there are full conferences that people do just thinking about how that interaction could be for different use cases, be it AI assisting in tutoring, ai assisting in medicine, ai helping pilots fly planes. 00:09:22.644 --> 00:09:31.046 It could be a whole range of different things and every one of them has slightly different priorities and therefore different interaction design aspects. 00:09:31.434 --> 00:09:35.802 And when you think about AI, human interaction design, you kind of want to think about four different things. 00:09:35.802 --> 00:09:43.274 You want to think about the human involved and what they bring to the table, and you want to think about the AI system involved and what it can do for the human. 00:09:43.274 --> 00:09:44.407 You want to think about the AI system involved and what it can do for the human. 00:09:44.407 --> 00:09:50.255 You want to think about the actual interaction modality how exactly is the human interacting? 00:09:50.255 --> 00:09:51.418 Is the human asking questions? 00:09:51.418 --> 00:09:53.884 Is the human just reading off of the AI? 00:09:53.884 --> 00:09:58.042 Is the AI requesting information from the human, and so on? 00:09:58.844 --> 00:10:03.664 But then you also have other contextual factors the environment and so on that the interaction is taking place in. 00:10:03.664 --> 00:10:08.894 All of this matters when it comes

to designing a good human interaction. 00:10:08.894 --> 00:10:17.970 And all of this is to say that it might be the case that you get results like this, where it seems like the doctors are performing worse. 00:10:17.970 --> 00:10:24.538 The confounding variable here could be that it's not that the doctor is necessarily getting worse. 00:10:24.538 --> 00:10:47.345 It's that the specific interaction design wasn't optimal for that purpose you could have an AI system working in real time with the endoscopist and you know, trying to identify these polyps. 00:10:48.015 --> 00:11:02.864 Or you could have an AI system, kind of like in the background, that's like passive or more of like a second reader, where you don't necessarily get the results right away, or you could have it act totally autonomously. 00:11:02.864 --> 00:11:14.687 But you're right, like now, that I think about it, it's not necessarily what is the AI doing, what is the physician doing, but actually, like, what is the interaction between the two that is so important? 00:11:15.414 --> 00:11:28.615 Yeah, and you have to remember that the physician has more information than the AI system, because the physician is not only looking at the images, but they're actually feeling around inside and they're actually, you know, sort of deciding where to go next and making plans about that. 00:11:28.615 --> 00:11:29.498 Reasoning about that. 00:11:29.498 --> 00:11:31.523 Those are not the things that the AI system is doing. 00:11:31.523 --> 00:11:49.559 For the most part, they're looking at images, right, and so if and if you imagine an interaction in which the AI and the human are looking at it together and there's things popping up on the screen and maybe this little square box is shown around certain things and given names, then that's going to offload that thinking for the human. 00:11:49.559 --> 00:12:02.897 Assuming that that is correct, right, but you can see the issue here, which is that that little boxing and labeling of that from the AI system is likely to be incorrect because it doesn't have all the information the human has. 00:12:02.897 --> 00:12:05.724 So now the human has to work with. 00:12:05.724 --> 00:12:24.664 It's interesting to explore this more deeply and I think that you hit the nail on the head in, the AI system could be concurrent with the human, and that's a scenario it could be a second reader where the human is encouraged to come up with their own analysis first before looking at an AI systems analysis, right? 00:12:24.664 --> 00:12:26.421 In which case it's not really. 00:12:26.421 --> 00:12:30.780 It's assisting the human in a not in an efficiency manner, but for an accuracy purpose. 00:12:30.780 --> 00:12:35.307 Maybe the human can discard what the AI is saying after the fact, right? 00:12:35.307 --> 00:12:39.547 Because if the human has considered things that the AI hasn't, that might be useful. 00:12:39.547 --> 00:12:42.623 And there are other ways to do this too. 00:12:42.623 --> 00:12:47.407 I mean, it could be the case that the AI's role is limited. 00:12:47.595 --> 00:12:52.206 The AI is only used for polyp detection versus polyp characterization. 00:12:52.206 --> 00:12:58.823 Maybe the AI is only supposed to do certain things and therefore you control the degree to which the human begins to rely on the AI system. 00:12:58.823 --> 00:13:15.144 And this gets into a really hazy area here, because again, we had the same similar sort of thing with episode one and diabetic retinopathy, but the difference there was that the accuracy of these systems was very high. 00:13:15.144 --> 00:13:27.164 The diabetic retinopathy AI systems were up in the 99.9% accuracy ranges, and so you could have the AI system, just look at a picture and tell you whether or not there was something bad. 00:13:27.164 --> 00:13:51.769 It seems to me here, based on the research, that the accuracy rate of these systems are much lower for polyp detection and, more importantly, the false positive rates are way higher, which means they're going to flag things that are not really polyps as polyps and or as I don't know, like the specific kind of polyp that's like pre-cancer as the adenomatous one. 00:13:52.195 --> 00:13:56.407 Right, and so then it becomes a question of okay, how do you build trust? 00:13:56.407 --> 00:14:00.962 How does the human build trust with the AI system in that scenario If it's producing false positives all the time? 00:14:01.043 --> 00:14:15.678 Okay, like I have a confession to make, though I don't know, I feel like, if AI like put a box around something and was like this is an adenoma, I like I might trust it more than I trust myself. 00:14:15.678 --> 00:14:22.198 Like, and that's probably not a good thing, right, that's a problem in this situation, and so I don't, like I don't really know what to make of that. 00:14:22.198 --> 00:14:24.325 Like, sometimes I do feel like, oh, it's objective. 00:14:24.325 --> 00:14:29.178 It put a box around it and it says that it's a, it's an adenoma, like we've got to take it out. 00:14:30.100 --> 00:14:38.668 Instead and maybe this is part of the AI human interaction that you design Maybe it just puts a box around it and says look at this one, what do you think? 00:14:38.668 --> 00:14:49.201 Instead of flagging exactly the type, it is right, because the more information it gives me, the more I'm like, oh, I believe exactly that. 00:14:49.201 --> 00:15:04.946 But if it's simply flagging it, you're creating a human AI interaction where you're saying, okay, I got to look at this one, but it's not telling me exactly what it is, cause I, I personally just like I assume it's correct if it flags it as X, but you're telling you know, and the research shows, and you're telling me it's definitely not always correct. 00:15:05.928 --> 00:15:13.315 Yeah, and you know, in support of the, in support of the study, you know, one point that could be made is that it's actually working right.

00:15:13.315 --> 00:15:19.919 Whatever interaction design they have going works because their adr rates or, I don't know, my detection rates, went up right. 00:15:20.120 --> 00:15:25.884 So, in a sense, with the ai, yeah, they went down when they took the ai away, but it went up with human plus ai. 00:15:25.943 --> 00:15:41.525 So you know, I would imagine that there might be other follow-on studies, um, in which they compare different interactional design patterns between the human and the AI and see which ones produce the best improvements in performance for the human AI team. 00:15:41.525 --> 00:15:43.341 Right, yeah, that's what you have here. 00:15:43.341 --> 00:15:48.004 You have here a human AI team and that's not the worst thing in the world, right? 00:15:48.004 --> 00:15:51.344 And so that's kind of what motivated the study in the first place. 00:15:51.344 --> 00:15:55.876 Now the question is they went back and they said well, the humans don't have that skill anymore. 00:15:55.876 --> 00:15:59.663 After three months, seems to have lost, seems their performance declined. 00:15:59.663 --> 00:16:13.344 Now it's unclear again whether the performance declined temporarily because they had just come off of using the AI system and they were used to that new paradigm, or it was there some actual degradation and memory loss. 00:16:13.344 --> 00:16:22.601 Memory loss is a strong word, but there's, you know, there's a bit of if you, you know you lose it if you don't use it, kind of you know kind of thing. 00:16:22.601 --> 00:16:24.009 Is that happening as? 00:16:24.028 --> 00:16:25.274 well, it's not like riding a bike. 00:16:25.274 --> 00:16:28.639 As it turns out, autonomic detection is not like riding a bike. 00:16:28.820 --> 00:16:29.240 That's right. 00:16:29.240 --> 00:16:30.101 It's not as internal. 00:16:30.101 --> 00:16:32.224 Maybe that's the case, we don't know that. 00:16:32.224 --> 00:16:42.467 Again, that's something we don't know from the study itself, but the study shows that, yes, that the humans do better with the AI system when they've been exposed to the AI system. 00:16:42.975 --> 00:16:43.395 But okay. 00:16:43.395 --> 00:16:51.424 So this is interesting to me because we have humans here that are trained, that have been doing this for a long time now. 00:16:51.424 --> 00:17:07.717 Imagine new trainees coming into the system, new doctors who are being trained for the first time in how to do endoscopies, and that's like a whole different ballgame, because now they're what it's kind of like the digital native version. 00:17:07.717 --> 00:17:19.444 They're AI natives, like they're coming into a system where AI is basically table stakes, right, and they're used to having these detection tools from the beginning. 00:17:19.444 --> 00:17:28.175 And this is interesting to me because we talk a lot about having a human in the loop and obviously there's definitely humans involved here, because there is human AI interaction. 00:17:28.175 --> 00:17:48.909 But this is so interesting to me because if you have humans who have only developed the skill with AI, how do you make sure that the adenoma detection remains good? 00:17:48.909 --> 00:18:01.827 Like what if the lighting were different or the camera change or something else happened to the AI system, would these humans be able to detect that? 00:18:02.107 --> 00:18:17.423 That's a great question and again I think I want to go back to the human AI interaction paradigm, which is this task of interaction design, where you're designing what the interaction should look like for different tasks of different purpose, different designs, your task in a real clinical setting. 00:18:17.423 --> 00:18:20.859 For a doctor trying to do this, it might be very different from a training session. 00:18:20.859 --> 00:18:23.844 In a training session, that interaction design might be entirely different. 00:18:23.844 --> 00:18:32.665 So when you can actually use an AI tool like this to help trainees by not giving them the answer right, by having to work with them, and then slowly building up the use case. 00:18:33.295 --> 00:18:33.816 So you train them. 00:18:33.816 --> 00:18:34.800 You teach them answer right, by having that passive slowly building up the use case. 00:18:34.800 --> 00:18:35.230 So you train them, you teach them. 00:18:35.230 --> 00:18:39.426 Okay, this is the thing you missed, or that is the thing you missed totally different problem to solve, but totally doable, right. 00:18:39.426 --> 00:18:41.715 And that's a different interaction design. 00:18:41.715 --> 00:18:50.000 And so you're not just giving them all the tools immediately, you're helping them understand first principles before they work themselves up to using the tools. 00:18:51.282 --> 00:19:00.420 It's really funny because you know, the example that I keep thinking about is I play a lot of chess and in chess you can play chess and computer systems are way better than any human right now. 00:19:00.420 --> 00:19:04.961 And people use computer engines, as they call them, to help them with their chess game. 00:19:04.961 --> 00:19:09.459 And the way they do that is they play the chess game and then afterwards they analyze their own game to look for mistakes. 00:19:09.459 --> 00:19:14.527 Now you can just turn on the engine at that point and the engine will tell you this move you made was really bad. 00:19:14.527 --> 00:19:21.280 It might not tell you why or how, it'll just tell you this move is really bad and it's correct. 00:19:21.280 --> 00:19:24.212 Now the question is it's up to the human to decide what to do with that information and they can move on with it. 00:19:24.212 --> 00:19:35.324 Now, as a trainee, you're encouraged to not use the engine to think for yourself and figure out what things, why that move was bad, and the way you do that is literally work through and work hard. 00:19:35.324 --> 00:19:45.909 You know you rack your brain and think about what was difficult

and what was wrong about that move, but by doing that once or twice you are now better equipped to not make that mistake again in the future. 00:19:45.909 --> 00:20:04.180 And so you could imagine a simple, similar training paradigm here as well, where you have trainees working without the AI system, training and learning the actual detection task and then utilizing the tool as almost like a superpower for them to be able to then do things faster and better. 00:20:04.180 --> 00:20:09.059 But it comes right back down to the specific interaction design. 00:20:10.263 --> 00:20:16.866 But with all that said, I do want to return to sort of an overarching question that we started off with, which is so what? 00:20:16.866 --> 00:20:17.650 Who cares? 00:20:17.650 --> 00:20:19.435 So there are these skills without that. 00:20:19.435 --> 00:20:21.124 I mean, we all use calculators now. 00:20:21.124 --> 00:20:22.390 We all use computers for things. 00:20:22.390 --> 00:20:24.057 We don't use typewriters anymore. 00:20:24.057 --> 00:20:25.942 Pilots use autopilot all the time. 00:20:25.942 --> 00:20:26.625 They rely on it. 00:20:26.625 --> 00:20:30.296 So humans do rely heavily on technology right now. 00:20:30.296 --> 00:20:32.260 And yes, of course you take all that away. 00:20:32.260 --> 00:20:35.567 We're not going to do things as well as fast as accurately. 00:20:35.567 --> 00:20:37.719 Is that such a bad thing? 00:20:39.605 --> 00:20:41.494 Yeah, I mean, that's a really good question. 00:20:41.494 --> 00:20:47.603 It's like we're developing new tools all the time, whether they're ai or not. 00:20:47.603 --> 00:20:51.193 In medicine, you think about, okay, people practice without having stethoscopes. 00:20:51.193 --> 00:20:58.878 Right now, I couldn't imagine walking around without a stethoscope, um, but yeah, I think that's. 00:20:58.878 --> 00:21:09.327 I don't have the answer to that question, but I think in some situations it could be okay to use ai as a tool, to depend on it in a way. 00:21:09.327 --> 00:21:20.060 When you've designed the interaction, that's appropriate, but you still need to be able to have oversight of what that interaction looks like and how that tool functions and make sure it's functioning properly. 00:21:20.650 --> 00:21:22.576 Yeah, and also, I think there are different kinds of tools. 00:21:22.576 --> 00:21:29.032 I think about stethoscopes sometimes, and I think that, before the stethoscope, I mean, how do people listen to heartbeats? 00:21:29.032 --> 00:21:34.436 Do they literally put their ears on someone's chest and listen to the heartbeat and listen to their lungs? 00:21:34.436 --> 00:21:38.039 You know, and that was clearly not efficient, not from a. 00:21:38.039 --> 00:21:43.723 You know, that was a better way, right, stethoscope was way better than doing that. 00:21:43.723 --> 00:21:46.965 It just amplified the signal that already existed, right? 00:21:47.165 --> 00:21:52.609 That's not what this AI is doing, though here. 00:21:52.609 --> 00:21:54.657 It's not taking your light image and giving you a clearer, better look. 00:21:54.657 --> 00:22:03.154 That would be a tool that would improve, give the human more information or more relevant information, but what it's doing here is doing some of the human's job. 00:22:03.154 --> 00:22:32.499 Right, it's doing the reasoning job of figuring out if something is a bad type of, you know, adenoma or whatever and that's a different type of task, and a lot of AI right now has been about automating human reasoning, and I think we should be a little careful with that, because it might seem like the AI systems are doing reasoning, but they're not, and it's, to some degree, that's where the over-reliance comes in, because you also, as a doctor, might feel like, oh, it's giving me a nice little box and giving me a very medical-sounding name for the thing Exactly. 00:22:32.719 --> 00:22:33.421 It must know better. 00:22:33.421 --> 00:22:44.778 And it's like taking the mental load off me, right, so now I can concentrate on just like getting the best picture or whatever it is, and like that mental load has been removed. 00:22:44.778 --> 00:22:54.971 It's like you basically like take that weight off and you concentrate on other things, which is not necessarily the best thing to do, right, right right, and it's a decision making step. 00:22:55.011 --> 00:22:58.099 Right, it's a step of characterizing, detection, decision making. 00:22:58.099 --> 00:23:02.523 It's not a step of receiving a signal and improving the quality of that signal. 00:23:02.523 --> 00:23:05.132 I think that's how I think about the differences between them. 00:23:05.132 --> 00:23:15.326 I mean, autopilot is another example where people you know airplanes, fly, fly themselves for the most part, and that's great, but there's always a human sitting there looking at everything, observing everything. 00:23:15.326 --> 00:23:17.478 It can jump into manual mode as needed. 00:23:17.478 --> 00:23:23.739 You know, you can imagine a system like that here, but I don't know that we're at that level of robustness. 00:23:23.739 --> 00:23:31.000 In these autopilot situations there's a lot of redundancy, there's a lot of guarantees about the performance of these systems and such. 00:23:31.000 --> 00:23:36.415 Here we don't have all of that, and so I think that it's a little different. 00:23:36.415 --> 00:23:45.284 In that scenario you have to really trust the system to do the right thing for the most part, and I don't think we're quite there yet with these sorts of computer-aided diagnosis. 00:23:46.390 --> 00:23:58.875 Yeah, I mean it sounds like we're not quite there yet, but there's a potential promise, um, for some of these systems, especially when there is this appropriate human ai interaction. 00:23:59.736 --> 00:24:11.123 Yeah, yeah so I mean, I guess it comes back to this question that I asked at the beginning, which is grammatically correct, I'm just pointing out is AI making doctors stupider? 00:24:11.123 --> 00:24:18.577 And I don't think we really have an answer, because what we unpacked in this episode is just that it's like not so simple. 00:24:18.577 --> 00:24:36.000 It's not so simple as the human and the AI, but it's also has to include the environment that you're in and the interaction between the human and AI, and that design and the task that at hand training versus doing the actual. 00:24:36.039 --> 00:24:36.280 Thing. 00:24:36.582 --> 00:24:39.218 Right, the design of that interaction. 00:24:39.218 --> 00:24:40.973 Yes, is key here. 00:24:41.214 --> 00:24:43.114 Yes, it can make all the difference. 00:24:43.114 --> 00:24:44.137 I think that's the big takeaway. 00:24:44.137 --> 00:24:53.442 So, you know, when people have read these stories, I think I would want them to think about not AI as this monolithic thing that just lives, but specifically about how it's being used. 00:24:53.442 --> 00:25:01.061 Because if, without that clarity, we won't know exactly what the result means, right, as we've seen here. 00:25:01.663 --> 00:25:04.115 Yeah, well, this is a great conversation. 00:25:04.115 --> 00:25:08.376 Thanks for joining us and we'll see you next week on Coding Cure. 00:25:08.376 --> 00:25:09.119 Thank you,