Latest / AI Meets MedTech / Preparing Hospital Data for the AI Agent Era - Eolas Medical
Transcript
- Anna Nesterova: Welcome to AIMets MedTech. Today my guest is Dr. Declan Kelly, the founder and CEO of OLAS, the knowledge management platform for healthcare organizations. Welcome to AIMITS MedTech. Dr Declan Kelly: Hi Anna, thanks very much for having me on. Great, great to be here. Anna Nesterova: Thanks for joining. I really have a lot of questions about your company, about the impressive journey that you've had so far with OLAS. But before I jump into my questions, I want to give you a chance to introduce yourself and tell my audience a little bit more about you and your company. Dr Declan Kelly: Yeah, perfect. No, thank you very much. so I suppose as a as a quick bit of background, I'm I'm Declan. I'm a medical doctor by background. I worked clinically as a doctor for about five years, mostly in the field of emergency medicine. but I'm also a software engineer as well. And essentially I I've always loved building things. And the problem that myself and my colleagues were facing when we were treating patients was we found it very difficult to access the knowledge that we needed within our own health system. So, you know, if someone's having a major heart attack, what's a referral? Pathway, what exactly do we do? You know, someone has failed, you know, third line cancer therapy, what treatments do we offer in this hospital and what's actually available to us, et cetera. Usually all that information either doesn't exist or it's in someone's head or it's on a SharePoint or an outdated intranet system. And so what we do is we use AI to index all of that content, and then within a few seconds, the clinician can ask a question and get an answer. And we launched that about four years ago. It's live now in about five hundred sites across the world. So about ninety percent of all acute hospitals in the in the NHS use it and there's about forty or fifty centers across the US, including places like Stanford, Mass General, Boston Children's and and a number of others as well. So it really seems to be to be making a difference and and helping helping clinicians deliver care. Anna Nesterova: Yeah, it is a massive success. And I want to say like in such a short period of time, you achieved such a large reach, which is very impressive. And I really want to kind of dive deep into the implementation side of things. But before I do so, I want to ask you, because you are both a doctor and a software engineer, as you said. So what was the moment of your career where you decided I want to solve this problem and I want to start a company and I kind of move away from the medical practice and I want to do this? Dr Declan Kelly: Yeah, it's a great question. And and it's it's one that I think gets asked more and more now because a lot of healthcare companies now are being founded by clinicians and doctors and and it's a problem that's kind of born out of frustration. And generally the answer is is quite similar to most of the answers actually that I've heard from those other physicians, which is I never really planned on this. You know, I never really planned on let's go and build a big business out of this as a problem. I think it was more of a This is a problem that exists, this needs solved. How can we use technology to solve it? And and doing that, and then realizing it's it's it's a ubiquitous problem globally, actually. and then it it it it's kind of snowballed from there. So I was very much stirring down the the you know the barrel of of 30 or 40 years working clinically and I loved it and I really, I really enjoyed doing it. And it was really just though the ability to build a solution and then see that solution get implemented and then the impact it had downstream and I think technology unlocks enormous gains in every industry and and healthcare thankfully now is not being spurred from that. So it was really the pull of of of that, the ability the ability to have a a difference and use technology to leverage that difference at scale that really then made me go down that path. So I I dropped my R's and I went clinically half time and then continued coding and did that and then it got to the point where I just I kinda ha had to choose and and then I went down down that path. Anna Nesterova: you you said it right. a lot of medical professionals are trying to make a difference and they do see how they can scale the impact with building something like what you're building. And can you tell me a little bit more about OLAS and the product itself? So you said that once you get implemented in the hospital, the doctor just asks for a point of information, but what exactly the system behind that product, like what does it do? Dr Declan Kelly: Yeah, great question. So a a good way to frame this as well and approach this is I suppose the whole the way the world's working now actually is pre large language models and post large language models and and and viewing the answer Anna Nesterova: Uh-huh. Okay. Dr Declan Kelly: through those two different lenses. because we we I started the company just before Transformers really properly hit the scene and and there was you know the the open AI chat GPT moment. So We initially were trying to index all of this messy textual data. So it's data, you know, in the form of PDFs, unstructured data, maybe it's tacit knowledge that lives in clinicians' heads, trying to find a way to represent that data. And we played about with, you know, graph databases and a way to represent that knowledge. And it was really only when then large language models hit the scene where then that data it becomes un unlocked essentially. so what we what we build is is a pipeline that takes any form of data at all, whether it's everything from From video rights down to to to documents, chunks all of that data up and indexes it and then makes it AI ready and then also human readable within seconds when you ask a question. So it's a rag-based system, you ask a question and get an answer. And the reason why I kind of framed it as a post, pre and post large language models is right up to this point, the only people needing The only entities needing to access that data, should I say, are are humans, you know, are you doctors, nurses, pharmacists, you know, asking a question and getting an answer. Now we're at the point where agents actually need access to that knowledge as well. So in your health system, if you're deploying an agent for triage or for, you know, diagnosis or whatever it may be, anything can now call on that, you know, vi via an MCP or via an an API call and then get that information that it needs back and then continue to make decisions. based on that latest kind of knowledge from your health system. So there's enormous opportunity now with unlocking all of that kind of unstructured, previously messy data that wasn't really being used properly. Anna Nesterova: Yeah, it's definitely some sort of a red tape that I see in my conversations with people who trying to solve the question of how to use the massive data in healthcare and how to implement it with AI tools and how to have a better use of that data with AI tools nowadays. And it seems like what you are doing is Preparing the data of internal data of the hospital to maybe go and work with a different company and different AI tool and kind of use that data set with a next tool. Is that correct? Am I understanding it correctly? I'm sorry if I don't go, but Dr Declan Kelly: Yeah, correct. So so Anna Nesterova: I'm just trying to to make the connection. So let's say you implement your product and your system in the hospital. So the next step is: okay, we want to add another tool with AI agent, like you said, or something else. We can use you. Your data, how you structure it and index it to build it up on. Is that correct? Yeah. Dr Declan Kelly: Correct, yeah. So I suppose really the the the the the two problems we solve is one side is ingesting all the knowledge that exists in your health system, and then the other Anna Nesterova: Yeah. Dr Declan Kelly: side is disseminating that knowledge to the point of care of where it needs to go. So we Anna Nesterova: Mm-hmm. Dr Declan Kelly: sync or ingest everything from your SharePoint, your intranet, all your policies, procedures, guidelines, all of that knowledge that Anna Nesterova: Okay. Dr Declan Kelly: lives maybe in people's heads. We bring all that into the system and then we allow it to be seamlessly accessible from anywhere. So within a a mobile and desktop app, so that the doctor can just ask a question. But also via their ambient listener, their voice technology, or their electronic patient record, or maybe any agents that they have running in their hospital or or will have running. So it's really that kind of two two two pronged problem that we're solving. Anna Nesterova: Yeah. And you said now you have your product in more than five hundred hospitals across the globe. So what has changed about how you think about AI and your product from hospital number one and now when you are in whatever number you are, so once once you scale the product, how how's your impression of your product changed? Dr Declan Kelly: Yeah, it's it's a great question. So I think the core fundamental change is who the end user is and where they're going to be using it is changing slowly and then it will change all at once. And and what do I mean by that? So the end user initially, when we made it for the first hospital, was ultimately myself and then my team within the emergency department and then the hospital. So we indexed all of that content using mobile app to get the answer. where that's Anna Nesterova: Mm. Dr Declan Kelly: now going is we actually will probably see less people going straight to the mobile app and the desktop app and they'll get it seamlessly within their workflow everywhere else. So you know in the electronic record, in the in the ambient voice technology. And then where it's going to go is get to the point where it's it's used by AI agents. So it's it's that that's the trajectory of change that we're actually seeing. And the as Anna Nesterova: Mm-hmm. Dr Declan Kelly: I say, that'll happen slowly at first. And then it'll happen then it'll happen very, very quickly. being able to tap into that knowledge base will start to more and more be happen happening by API calls and and and Anna Nesterova: Mm-hmm. Dr Declan Kelly: MCP. Anna Nesterova: And what was your implementation process? What are challenges that you faced going from side to side? Because it's one thing to implement it with your own team, as you said, because everyone knows you, everyone understands the challenge, but was it a hard process for you to transition and convince other sites to try and use your product? Dr Declan Kelly: the short answer yes. I think trying to, you know, trying to get anything Anna Nesterova: Yeah. Dr Declan Kelly: implemented in and adopted in healthcare Anna Nesterova: Yeah. Dr Declan Kelly: is extremely difficult. we're we're Anna Nesterova: Yeah. Dr Declan Kelly: now seeing that change, you know, we're seeing product led growth happen for the first time really properly in healthcare. And you know, we're seeing it with external answer engines like open evidence and we're seeing it with ambient voice technologies, with you know, like a bridge and lots of the other ones are now being adopted at the clinician level. Once I think If you're able to really deeply solve a problem for a doctor and make it really easy for them to use with no blockers from their system, it then starts to spread very, very easily. And you have to figure both of those things out. otherwise it has to be a traditional healthcare seal, which is a top down sale, you know, going to a large Anna Nesterova: Mm-hmm. Dr Declan Kelly: health system, talking to the leaders, working with them for 18 months, aligning with their strategic priorities and then selling top down. So we we do we do both of those, where we have the ability for any individual doctor to use the app just straight away. but then we're now more working on with systems top down. So the the ability to do both of them is really key to to getting distribution essentially. Anna Nesterova: Yeah. And it is a hard process and I I lived that myself, so I understand and especially sometimes going top down. I feel like healthcare professionals can feel neglected. Have you seen that happening? Or they were pretty easy to convince to use your platform. Dr Declan Kelly: Yeah, so I think this is a it's a great point. And actually I I'm doing a webinar tomorrow on this very topic, you know, why is AI adoption so hard? and I think it's the explanation for that is any tool has to take into consideration if it's going to be successful, this idea of the end user of the clinician and what problem does it solve for them. as well as what Anna Nesterova: Uh-huh. Dr Declan Kelly: problem does this solve for the actual hospital health system and what metric does it move? And I think if we take you know electronic patient records, for example, you know, Epic and and a lot of all the other ones, I think if you survey or ask any physician in in the US, you know, do you love using Epic every day? Like most of them will probably say no, but it's not for them. It doesn't solve, it's not for solving their problems, it's for solving the problem of the health system with regards to building, et cetera. And that's how it's built. And the for tools to be successful now moving forward and and start to displace or move into this ecosystem, both of those things need to be true. You need to solve the problem for the end user so it gets adopted, but it also needs to move a number that matters for the health system. So does it increase the efficiency of care? Does it increase the quality of care? Does it decrease the cost of the care? does it allow more patients to be seen? so so you have to actually prove that that works. And the ambient voice technologies are a perfect example of that where it's shown to in decrease burnout, increase the efficiency of care, allow more patients to be seen, et cetera. And that actually moves the needle for the health system. So for the first time, we're kind of seeing these technologies really in use that the clinicians really want to adopt and use, but also it maps to exactly what the healthcare executives want to see as well, which is you know, better care for patients, but more efficient care, more affordable care. And Anna Nesterova: Yeah. Dr Declan Kelly: and and numbers that matter to that. Anna Nesterova: Can you tell me a little bit more about ambient? Dr Declan Kelly: yes. So so the ambient voice technology So that would sit in the room with the the doctor and the patient and it would listen to the transcript and then it would transcribe it and then they can start to maybe take actions downstream of that. So that that is, you know, probably the the the example of the most successfully adopted clinical AI use over the last probably five to ten years outside of stroke you know convolutional neural networks for for stroke detection. so it's been an enormous amount of investment obviously and now most health systems in the US are are using ambient voice technology. Anna Nesterova: Yeah. I Actually read an article a couple of days ago where even with that technology there are some challenges. And I know some I think it was a hospital in Boston that shared their data that in emergency departments, for example, it didn't save that much time as the hospital hoped for for the clinicians. Dr Declan Kelly: Yes. Anna Nesterova: And they're starting looking into bottlenecks where the problem really sits in the hospital. And I think it's it's a long process to discover what needs to be changed Dr Declan Kelly: Yes. Anna Nesterova: to run it smoothly. So what do you think about that? Like what do you think If we're talking about future, not specifically OLAS, but in general, with these new technologies coming to play and being adopted, do you think we all are super positive and thinking things will go smoothly, or do you expect having some sort of backlashes and resistance happening as we go and as we study them? Dr Declan Kelly: Yeah, no, of course yeah, there there will there will almost certainly be setbacks and you know, backlash and resistance and be a lot of good that will happen and then there'll be some bad that will happen. I think ultimately as a as an industry, evidence is obviously extremely important and it's extremely important that we show that these work and we actually measure what works. but in context of that, there is maybe a few important points. the the absence of evidence doesn't indicate the evidence of absence. so just because Anna Nesterova: Yes. Dr Declan Kelly: there's no evidence that it improves this doesn't mean that that's not true. I would also say that evidence is generally in with these technologies is a lagging indicator. it's it's funny the this is the first time that medical device like technology. So the they're s ultimately their software is as medical devices, even though you know the FDA hasn't hasn't actually said that yet. they're they're software that that's working in a very high risk environment. And it's for the first time the model of how to validate and get evidence around if that's useful is completely different than how as a society we've decided to to test those things. And what I mean by that is historically the only thing we can compare this to is pharmaceutical industry, so drugs. and medical device industry, so physical medical devices, they don't get into clinical practice until they go through rigorous testing and evidence based trials, phase one, two, three. That takes a decade sometimes for some drugs and medical devices. And then they get launched into into widespread use. The complete opposite has happened with these the software technologies. So, you know, open evidence, for example, is now used by eighty percent of all physicians across the US. Ambient listening is used by roughly around eighty percent as well. And that has been adopted, you know, at scale before rigorous, widespread evidence can actually be generated. And whether that's a good thing or a bad thing is, you know, up for Anna Nesterova: Mm-hmm. Dr Declan Kelly: a large debate that that'll take a couple of hours. But I think that's the position that we found ourselves in. and we now need to assess these technologies, prove that they're safe, prove that they do Anna Nesterova: Yeah. Dr Declan Kelly: much more good than harm, but also prove that they actually in improve patient care as well. So we're I what I I what I think will happen is is over the next number of years, all of that evidence will start coming out and there'll be a number of landmark trials that will show this. and there will be trials that show this doesn't change or or move the needle with regard to the number of patients seen, et cetera, et cetera. But a lot of intangible benefits are also there as well. so yeah, Anna Nesterova: Mm-hmm. Dr Declan Kelly: there be I think there'll be a lot of interesting studies over the next three or four years around these technologies. Anna Nesterova: Yeah. Yeah. And actually using your position of being both medical doctor and software engineer who builds the platform a for AI, what can you tell potential patients who have skepticism around AI use being used in healthcare? Because of course when people see news, and of course we all have negative news around AI, especially in the last couple of days. What can you tell to them to reassure that what's happening in healthcare and AI is actually safe for patients and for the better cause versus, you know, you have to be scared and skeptical. Dr Declan Kelly: Yeah, yeah, it's a great question. I think I think actually in general as there's a a lot of different ways you know, to approach that as a question. I think ultimately in general what patients want is safe and efficient care. And the filter for that care to date has always been healthcare professionals, you know, physicians, clinicians, pharmacists, nurses, et cetera. And that's still true today. I if a if a doctor doesn't think that an ambient listener is gonna make that doctor safer and better, they're not gonna use it. and they're not gonna implement AI in their practice if they think it's gonna harm their patient. And so, you know, medicine over the last number of decades has has swung between this kind of very paternalistic, which is which is the wrong way to be, but and then very patient-empowered. And we need to get a combination of both. The doctors are still the experts. They don't want to do any harm to their patients. They want to do the best for their patients and they're there to make sure that that happens. And so if a doctor feels it's appropriate to use AI safely in their practice, that generally is a pretty good filter. No doctor is just going to randomly trust Chat GPT and just and just go off that as an answer. Yes, they're gonna use it in their practice, but they're gonna use it alongside everything else with their clinical gestalt and all of the systems that they have in the health system. It's not just a case of ChatGPT told me do this, therefore I'm gonna Anna Nesterova: Mm-hmm. Dr Declan Kelly: do this. so I I would say I'm more optimistic currently. Anna Nesterova: Yeah. Yeah. Thank you. And I actually have a question about the difference in mindset between Europeans and Americans. Have you seen the difference how healthcare professionals and the executives of health systems in the US react to your product versus people in Europe? Is there a difference at all? Dr Declan Kelly: Yeah. there is an enormous difference. Yeah. Anna Nesterova: Ha ha ha. Dr Declan Kelly: The risk appetite in the US is is much, much higher. the instinct to regulate is much higher in the UK and is much higher again in Europe than it is in the US. Now obviously lots of things are very highly regulated in the US, but when it comes to healthcare, specifically healthcare software in the US, it's it's a lot more favorable. for innovation to occur and to allow solutions to get to market whereas in in Europe it's very, very heavy on the regulation. Anna Nesterova: Yeah. I think all us is doing something right if if the UK decided to implement your system and allow you to actually i use their data sets. I think you're definitely Dr Declan Kelly: Yes, no, exactly. Yeah, yeah, yeah. Anna Nesterova: doing something right. And you've spoken about the Golden Triangle framework. Could you please tell me what it is and why it's important when building AI products in healthcare? Dr Declan Kelly: Yeah, so I I probably could have come up with a better name for it in general, but it was really based upon the principle of when a healthcare professional is seeing a patient, you know, three things are happening at one time, and those Anna Nesterova: Mm-hmm. Dr Declan Kelly: three things can now all be done by an AI system and and that's extremely powerful. And I actually think that will set the roadmap for how healthcare is going to be delivered in the next, you know, ten, twenty, thirty, forty years. And it actually sets the roadmap and our ability to actually start to think about building autonomous health systems. So, what are those three things that a doctor does when they see a patient? So generally they'll see a patient and they'll look through all that patient's history. So that usually is in the patient chart and they'll review their entire chart. Then they take the history from the patient, i.e., you hey, why are you here today? You know, where's your chest pain, etc.? They'll examine the patient. So they'll see what's going on at that point in time. That's now being captured. By the ambient voice technology or the scribes, you know, automatically in the room. And then what they do is they we say, okay, for this patient, you know, I know everything about this 60-year-old man and I know everything that's happening right now. I'm going to use those signs and symptoms and tests and query a knowledge base to see what the next best step is for their treatment. And generally what we query before, you know, like 10 years ago, it was maybe querying our heads or looking something up on Google or looking something up on you know up to date or one of those platforms. And now we're able to seamlessly use AI to query the world's latest medical knowledge within seconds. And so and both internally, which is what OLIS does for your health systems knowledge, and then externally, which is what you know up-to-date open evidence and lots of other ones do as well. So we're able to take those three things and they're all converging to one point, which is the ability to make an autonomous decision for a patient and base that decision on the latest evidence that exists in the world and within your health system. and make it hyper personalised to that patient based on all their signs, symptoms, medications, et cetera. So th that's an incredibly powerful moment Anna Nesterova: Yeah. Yeah. Dr Declan Kelly: and it is a huge potential downstream. Anna Nesterova: Yeah. But I think we need to achieve that liberation of data from the healthcare systems in order to achieve the real results, like I guess of of what you're describing. Is is that correct? That's what do you think too? Dr Declan Kelly: Yes. Th this is a a lovely picture and vision of this triangle, but the problem is each each of those areas of the triangle are extremely messy. You know, if we look at the electronic record, it's extremely unstructured Anna Nesterova: Yeah. Yeah. Dr Declan Kelly: data. It's very, very difficult to extract anything. I think the Anna Nesterova: Yeah. Dr Declan Kelly: the if we if we look at this as a point of time, everything before ambient voice technology was recording the history, all of that data is extremely unstructured and it almost you know, it's it's not impossible Anna Nesterova: Yeah. Dr Declan Kelly: but Very, very difficult to work with. Now Anna Nesterova: Yeah. Dr Declan Kelly: we're getting to the point where if every single patient consultation is automatically recorded and structured in such a way that actually makes sense downstream of this over the next 20 years, all of that data is unbelievably valuable. And this will usher in this exponential age of clinical AI Anna Nesterova: Yeah. Dr Declan Kelly: because then we will have 30 years of data that is extremely well structured, captured, and barely anything is missed. Yes, there's going to be errors and it, yes, there's going to be mistakes and stuff on AI isn't perfect, of course. But it means we're moving from, you know, 20 years or 30 years of really unstructured, really messy data to then 20 years of, you know, perfect data almost for every single consultation. And the the value downstream of that is is enormous for patients and for the healthcare industry in in general. But you know, the amount of the the ability to do research on that data, the ability to find cures, the ability to actually treat treat patients patients better and and not make mistakes with regards to diagnosing etcetera, th the the opportunities are enormous. Anna Nesterova: Yeah, well that sounds fantastic and I hope we we'll we'll get there in in Dr Declan Kelly: Yeah, yeah, yeah. Anna Nesterova: a in a most efficient and harmless way. That's my hope is. Yeah. Dr Declan Kelly: Exactly. Yeah, of course. Yeah, yeah. Anna Nesterova: And it I I I like to ask my guests a question about something that surprised them recently in AI and healthcare. Is there anything that surpr genuinely surprised you recently? Good or bad? Dr Declan Kelly: Hey Yeah, hey I would say the speed of adoption is is is probably surprising given the the reticence for the h industry to adopt software or technology in general. It's usually quite slow. If we look at, you know, pr kind of pre-large language models, you know, the every industry had a SaaS platform for something. You know, a a accounting had all the best SaaS platforms and and finance and and and law, et cetera. They all had good platforms. Whereas healthcare generally, like we still were on paper, for example, over the last couple of years, we had paper patient records and a lot of systems still do. So healthcare generally is very slow to adopt new technology. Whereas now we've gone from a world of, you know, generative models or transformer models weren't readily accessible to within three years, eighty percent of all physicians in the UK are using or sorry, in the US are using one. to actually make decisions for patients. You know, I think the speed at adoption is is surprising when you're living it and and actually building in it and using them yourself day to day. It's not surprising 'cause the tools are amazing. but actually if you zoom back out and and look at it like kind of a fifteen year view, this the speed of adoption over the last number of years compared to the the ten years prior to that have been have just been day and night. Anna Nesterova: And do you think the large language models are the reason why everyone is trying to stay on the wave of that new tech, new era of AI? Do you think that's the main reason? Or maybe there's another one that you think of? Dr Declan Kelly: probably that is the catalyst. I think that's eighty percent of it. Anna Nesterova: Yeah. Dr Declan Kelly: I think the the the twenty to thirty percent of it is the fact that healthcare is ultimately In a very difficult position globally. you know, there's never going to be enough doctors, nurses, pharmacists. Anna Nesterova: Yeah. Yeah. Dr Declan Kelly: if the demand for healthcare is only increasing throughout every single country. And I I think that demand is impossible to service if we don't truly leverage AI properly. So it's come at a time of of kind of desperate need for for health systems. but yeah, I think it's been catalyzed by the technology just like it has been in every other industry. Anna Nesterova: And have you had a chance to work with governments? Because obviously NHS is a government funded organization. Because I'm Canadian, so I also lived in and worked in a similar healthcare system. So what was your perception of how government is thinking about AI and adoption of in healthcare? If you had a chance to work with them. Dr Declan Kelly: Yeah, yeah. So no, so we do we we we do and I think especially in the UK given the penetration that we have across all the health systems, we know, you know, in real time what every doctor's searching for, what they're accessing, what drugs are being recommended, which drugs aren't being recommended, et cetera. So that data is very, very useful at a at a government level, or it's very useful at a at a health system level as well. so so we do and we have these conversations. I think probably probably twofold. The first one is Not even related to healthcare, but just more related to the economy in general. I think again, globally, many economies are in in a difficult spot, you know, other than the US at the minute because of the AI bubble essentially. But lots of other economies are gonna really need to be reliant on on AI to try to boost GDP and boost productivity. So what we're seeing is them leaning into that very, very heavily, you know, giving out Anna Nesterova: Mm-hmm. Dr Declan Kelly: lots of grants to try to stimulate growth in this area so that it actually is able to then stimulate GDP. with regards to healthcare in general as well, healthcare costs are increasing. And if you if you're working in a system, you know, in in the UK for example, about 82% of all care delivered in the UK is government funded. That's kind of slightly going down and private care is kind of slightly going up. But ultimately the majority of the care is is government funded and it's only getting more and more expensive to deliver care. So they're very much looking at AI of how can we decrease the cost of care, how can we make it more accessible? And how I think the other thing that I do a bit of research on is how can AI ensure deflationary economics in healthcare? generally technology in every other industry results in deflationary economics. So you get better products for cheaper. but Anna Nesterova: Yeah. Yeah. Dr Declan Kelly: the opposite happens in healthcare. gen if you look at the, you know, the chart in the US, you know, the more spend and and worse outcomes. And that's a trend that that AI is very well positioned to to to upend. And I think a lot of governments are thinking through that lens as well. how do how do we ensure deflationary economics when we're delivering healthcare and AI is perfectly placed to be able to do that. Anna Nesterova: I want to finish our conversation with a short blitz. I prepared a few questions for you, just to give me a short answer. what do you think Dr Declan Kelly: Correct? Anna Nesterova: about them? So let's do that. So medical degree or computer science degree? Dr Declan Kelly: a medical degree. Anna Nesterova: I would love to know why, but we're gonna go for short answers. The biggest thing you got wrong about healthcare before you build your company. Dr Declan Kelly: Understanding how the system works from a top down perspective as opposed to just one per perspective, which is the physician's perspective. Anna Nesterova: Mm-hmm. one technology that will change healthcare in the next five years. Dr Declan Kelly: A Anna Nesterova: one thing AI will never replace in medicine and healthcare. Dr Declan Kelly: the the human to human connection and empathy that patients actually need when they're in the worst time of their lives. I think there's been doctors for four thousand years and there will be doctors for another four thousand years. The tools that they Anna Nesterova: Exactly. Dr Declan Kelly: use will be different, but there will always be that human to human connection and that ultimately is what at the core of being a healthcare professional is. Anna Nesterova: Yeah, and will AI make healthcare more equal or more unequal in the long run? Dr Declan Kelly: In the very long run, more equal. In the short term, probably it will increase inequality. Anna Nesterova: Yeah. Thank you so much for being on AI Meets MedTech. Could you please tell me and my audience how they can find you and contact you and your company? Dr Declan Kelly: Yeah, of course. so LinkedIn is probably the the best place for myself. So Anna Nesterova: Mm-hmm. Dr Declan Kelly: just find me on LinkedIn, reach out to me, send a connection Anna Nesterova: Sure. Dr Declan Kelly: request, more than happy to to have a chat. Anna Nesterova: Awesome. And of course I'll leave all the details in the description of the video. But thanks again for being here. Thanks. Dr Declan Kelly: Perfect. Yeah. It's great. Thanks, Anna.