Ai in Healthcare: The future is now! #aiimpact #aiinhealthcare

Introduction to AI in Healthcare and Its Potential Impact

[00:00] Is it telling you whether the risk of diabetes? There will come a point where maybe the learning will be so deep for the software, the AI, you won't have to check it.

[00:20] I think it's probably going to get better. In healthcare, regulations, ethics, all these things are so important, right? So you just have to go above and beyond and make sure that you're putting all the right systems in place. With so much of data coming in, AI is going to be a fantastic game changer. Then efficiency of doctors, that's where your team and your

[00:40] companies working. We now know that the accuracy of extra diagnosis with an AI is much higher than pure radiologists put together. It's a fact. There's a study that looked at this and they looked at all the data that has been put into an AI and now the AI can tell you what race the person has from whom the x-ray has been taken, which a doctor can't.

[01:00] Cardiologists will not be able to look at an ECG and tell you what is your risk of diabetes, but an AI can now. There are many possibilities. It's what we don't know and it's still there that's scary at times. We don't know exactly who the winners are going to be for radiology reporting where you don't even have to

[01:20] dictate the entire report at all. You just give a few findings and the report will get generated. 10x faster. Completely believe that, like you were saying yourself, that people who use AI will probably be the ones who survive. But people who don't use AI

[01:40] What I'm going to do today is have a demonstration of a clinical scenario. Me as a doctor sitting in a clinic with my patient in front of me and I'm going to demonstrate something that might surprise you all but in a good way. So as a doctor if I spend one hour with a patient I end up spending two hours documenting

Challenges in Clinical Documentation and AI's Role in Solving Them

[02:00] and doing administrative work. And this is a big no-no and in the future this is going to change. And people who help us change this and who are developing softwares and AI tools that are going to help us achieve this, I have one major player in that field, Mr. Rustam, who is the founder of Ognito sitting in front of me. But before I let you talk,

[02:20] I will have to demonstrate this to my viewers with your permission. So before we go ahead, so I am the doctor. Mr. Rustom is the patient and we are going to have a scenario for three minutes and we'll take it from there. So Mr. Rustom, welcome. What brings you into my clinic today? I'll just switch it on. Please.

[02:40] Okay, we're ready, we go. Hi, Ristam, how are you? Doing good, doctor. Just feeling a bit feverish and a bit of a sore throat and feeling quite swollen in my throat. I thought it just landed. Yes. Right. And since when are you having this problem? I've been travelling a lot, so feeling a little run down, but about three days. Any fever?

[03:00] Yes. Have you measured it? How was it? I did. Last evening it was about 101. Did you take something? I took a Panadol. And temperature came down? It did come down but came back in the morning. Do you have any chills or rigors? Body ache. Did you travel anywhere other than India yesterday? The week before I was in Delhi. Anything unusual?

[03:20] Do you have diarrhea or chest pain or anything unusual going on? Nothing like that, but I think just the sort of, you know, little bit of a horsey sort of coffin. Right. And are you on any antibiotics? I'm not. Right. And do you think you're getting better in the last 24 hours or it's getting worse? No, I'm still feeling quite low. Okay. What I'll do is what about a background?

[03:40] Are you a smoker? Do you drink? No, I'm not a smoker. Nothing at all. I mean, I have an occasional drink. Anybody else in your family ill? Just diabetes in the family. Okay, no other significant issues. Not significant. And past medical history, are you taking anything medications or anything? I am on statins. Okay. And how long have you been taking them? Nearly 10 years. Right. And you're okay with that? No.

[04:00] complications. And no significant family history, you said? No. Any allergies you are suffering from? I am actually allergic to the antibiotic. And what happened to you when you took it? So my throat starts sort of closing up actually. So I'm assuming that the last paracetamol you took was a few hours ago? Yeah, it was last evening. I'm just going to throw

[04:20] examination at you, just check your abdomen and listen to your chest and your throat and we'll take it from there. So I'm examining you now, just imagine I'm examining you now, I'm taking your throat. So your throat looks red to me. There are some pustules on it, I can see. Let me just listen to your chest. Your chest sounds clear, heart sounds are normal. I'm feeling your abdomen. There is no against

[04:40] an emegaly, no tenderness and it's soft. So I can see that you've got some redness in the back of your throat and some white spots and I'm thinking that you have an upper respiratory bacterial infection. So what I'll do is I'll take a throat swab, I'll do some blood test on you and the results will come in a few hours and I'll give you the result and I will put you on some antibiotics if needed and I will

Live Demonstration: AI-Generated Clinical Notes in Real-Time

[05:00] short after. So we'll arrange a follow up appointment in 48 hours with you. Sure. Okay. Thank you. Thank you very much. So that's the end of the conversation. So I can put a stop to that listener on the mobile phone. So that takes about 30 seconds to process. Okay. So while I'm challenging you,

[05:20] riding away with my patient and saying okay so we will meet after 48 hours nothing to worry it looks all straightforward and the antibodies are going to help him. My mobile phone here is kind of trying to process something and in 30 seconds it will be ready. So yeah it is already ready. Okay. Yeah here you go. So I am done with him he has walked out of

[05:40] side though and there you go. I have my medical report categorized the way I want it with chief complaints, allergies, history of present illness, past medical history, social history, family history, physical examination findings, diagnosis, lab test and plan of care laid down in a methodical manner. All I do is I

[06:00] press send and this report goes off to the patient or to the EMR or medical report storage that we have in our hospital. And that is what AI has done to documentation and reducing the time that doctors spend with patients. So now let me introduce this fabulous man who is

[06:20] come all the way from India. He's been here in Dubai for quite some time, but he travels up and down. So Mr. Rustam, welcome to Dubai now. And you're probably the first AI guru on an Oscar of Dentsuits. Great pleasure to be here. Thank you. Would you like Dubai? Oh, I love Dubai. I love Dubai. So that's why we're very, very focused here. And our strategy is very

[06:40] focused in on the GCC region, but Dubai is very much central to that strategy. And we actually got our first contract about a couple of years ago here in Dubai, but really ramping up in the region now. So before we go into the core work, which you've done, which is fantastic, I probably won't have enough time to discuss it, but we go

[07:00] as deep as we can in that. Tell me about yourself. So I know that you kind of have a bit of an experience in many parts of the world, including Europe, many countries. Just tell me about what you've done. Yeah. So I'm a really a health tech entrepreneur because this is now my second startup in health tech. My first company was called ScribeTech, which

[07:20] essentially was a medical transcription, business process outsourcing business which was focused on providing efficiency to the National Health Service in the UK, to the NHS. So that was a straightforward speech to text software. So this was in the early 2000s. So the way that that would work is we provide a software called digital dictation.

[07:40] software and the doctors would dictate on a digital voice recorder and then that voice file would be allocated to someone in Bangalore who's sitting there, would listen to the file, they would type it up, then it would go through a quality checking process and then once it's through that, it would

Understanding AI, Machine Learning, and Deep Learning in Medicine

[08:00] then be sent back to the digital dictation system which would then interface again with the EMR and put the letter into the hospital system. So in a way, I've been sort of working on this area of clinical administration for many years and it was really during this period that

[08:20] I was fascinated with automation technology and it was really around 2012 that I started investing quite heavily in R&D and we had actually worked with a professor from MIT and we developed a clinical speech recognition engine just as well.

[08:40] early years when deep learning was starting to sort of come through and we developed this technology and what we were able to do is really transform our transcription business because what we did is the file would still come to us, it would run through a speech recognition server and then someone would edit that file and that was the sort of

[09:00] improvement engine that had started with this technology and that reduced turnaround times for the service that we are offering because we used to do 24 hour turnaround time, now we could sort of do it in 2-3 hours. But if I take you one step back, that's when the other software is where the automatic speech to text without

[09:20] middle person being there came into existence. Exactly and it's really the deep learning technology, it's when the deep learning technology really kicked in around 2012 that it started becoming very good. So I think in the past if you got 70% it's better to type it frankly.

[09:40] waste of time having something that you have to edit so much. So it's only when it started getting to the sort of 90% type of levels that it started becoming efficient. Yes. For our audience, you have to tell what deep learning is. So I'm going to hold on to you for your little boil there. So deep learning is based on essentially what you call

[10:00] neural networks. Now the neural networks were essentially, they had looked at how humans have neurons and stuff and they had designed this based on how the brain works. So this AI technology has actually been conceptualized since the 1950s. It's just that we didn't have the compute power that

[10:20] was needed to make this work. So deep learning is really when the GPU platforms and all the sort of GPU cards and all started kicking in and what we found is if you could put in a lot of trained data, it would become as good as a human in a particular task.

[10:40] Even better in some areas. So say for example, typing a complex clinical report, now you had speech recognition which could do something like that. So that's really what deep learning's capabilities were. But today now it's more sort of generative AI and these type of things, which is a bit different because this technology now

[11:00] doesn't only do what it's sort of set to do, it can actually do things and generate new type of content, new types of images, etc. So it's a more capable type of system and it has the sort of ability to sort of think in a way. So we will come onto that in more detail.

[11:20] For me, for me, I'm a doctor, I go to a few hospitals, I work in my own clinic and my main issue is the time I spend with the patient. So I'm in my clinic, I'm on a keyboard in front of a computer typing away as the patient speaks to me. We've lost that connect with the patient. So I'm not looking at, I probably would spend 10 seconds in that 20 minutes or 15 minutes

[11:40] consultation that I look at the patient or speak to the parent in my case. But I'll be on this keyboard typing away while I'm listening to them. That's not great, isn't it? This is not something that can exist for very long and your patient might not stay with you for long if you don't have any relationship with the patient. And what this is doing, a software like this is saving me that time and effort that I'm putting in, not just that

Transformative Role of AI in Diagnostics and Decision Support

[12:00] The transcribers time and the people who are employed to do that and the time it takes for the whole process to finish. And it's also context switching. If you think about it, I have a, in your case, a youngster who's there and you have to give you sort of all your attention there.

[12:20] Then you have to context switch by looking at a screen and then get back to the patient and then get back to this. So I think one of the core design principles we had when we were developing this is that at the end of the day, you can focus completely on the patient and there's something that's invisible. It's actually listening and doing its job.

[12:40] Like you would have a very sophisticated scribe who's sitting there and just getting this done for you. How accurate is it? It's very accurate. It's very, very accurate. It's in the above the sort of 95% sort of capabilities today. So what that really means in practice is that

[13:00] is that yeah, maybe there's one or two edits that you would need to do in a particular file. As of now. But of course, I mean, we've seen how quickly these things are improving and we really do believe that it's it. So, when the report goes out to the patient now, I think many the hospitals that are using it, the clinicians that are using it do mention that this has been AI generated but edited or approved by AI.

[13:20] doctor at the end. Yes, I mean it would be that's in the sort of terms and conditions. But there will come a point where maybe the learning will be so deep for the software, the AI, you won't have to check it maybe. I think it's probably going to get there. But I think in healthcare regulations,

[13:40] ethics, all these things are so important. So you just have to go above and beyond and make sure that you're putting all the right systems in place when you do this. So let's go to the basics now. Let's start from AI. So AI for doctors, it's what internet was 30 years ago when internet came in and we doctors ran away from it to begin with. Eventually we realized we can't live without it. So we

[14:00] We have to learn internet usage. I think we are at that stage with AI now. I was speaking to some of my colleagues. I just released a video about it as well where we've pointed out to the fact that most of the doctors don't even use a straightforward AI tool like ChatGPT at this point of time. Some do, but many don't. Now, can you tell me what AI is?

[14:20] in just a brief, simple explanation. Yeah, I mean, AI is essentially being able to mimic intelligence the way that humans have had intelligence. So it started off really being able to do very specific tasks. And now we're moving towards a more generalized type of intelligence, not a complete generalized. Today we're

[14:40] they talk about like AGI which is. So you are talking about unimodal and multimodal. So moving on to from AI to the machine learning. So something like a chat GPT now that is a multimodal LLM they call it is not it. Large language models. So that is a step

[15:00] forward. It's a step forward just because the amounts of data that have gone into this are massive and these models are trained really to be not just deal with a specific task, it can deal with multiple tasks. So it's more generalized in that sense. So these LLMs are like chat GPT or chat GPT-4 now. They

[15:20] We are akin to generative AIs. Yes. Generative AIs. This is very much based on what they call the transformer models. Yes. So they can create something from just instructions. Exactly. Okay. And the instructions that we as users give, I have to learn that. Isn't it how to give that instruction? So they've I mean it's today.

[15:40] very famous sort of new area called prompt engineering. Prompt engineering, yes. Prompt engineering is definitely a skill. So me as a doctor, I want to learn prompt engineering. What should I do? I think today there's so much that you can learn online. I think you can go, you can take courses and I think the best way to probably learn prompt engineering is to start getting your hands

Future of AI in Medicine: Personalized Healthcare and Monitoring

[16:00] start playing with it. And to begin with, a chat GPT on your mobile phone is what you can start with. Exactly. And I think the limit to this is endless, isn't it? We don't know where this is going to end up eventually. I certainly don't know where it's going to go. I mean we can hazard some sort of game

[16:20] But I think it would be unfair to really say that we know where it's going to go. But I think at least in the sort of short and medium term, there is a view where this is going to go. And I think it is very, very powerful. Especially when you look at healthcare as an industry, I think this is really one of the industries that

[16:40] massive amounts of improvement coming from this technology because you have doctor-patient ratios, you have nurse-patient ratios, all these type of things that how do you make clinicians so much more efficient through these type of technologies?

[17:00] sort of super-staffing in a way that you can imagine today you have speech to text but you have also TTS which is text to speech. So today say for example if you got personalized healthcare plans which are being created by AI for example you are supposed to take your medications at a particular

[17:20] Now today with these technologies, the TTS can give you a call and say, hey, have you taken your medicines on this time? Now you're in the past, you may have needed call centers to do that, etc, which was never really scalable. But this is truly scalable. So I've been doing my research in AI and healthcare and I've developed an acronym for this which is called DREAM.

[17:40] When we break it down, D is what we call diagnostic accuracies. In terms of how far air has gone with helping diagnose things more accurately, we probably don't know that as much. A few examples in that D we will be talking about later.

[18:00] research and I'm sure you know that in research with so much of data coming in, AI is going to be a fantastic game changer. Then efficiency of doctors, that's where your team and your company is working. The other is early detection. So before I have Alzheimer's, would I know I have Alzheimer's? If there is a biomarker that can tell me, that will be great. So I can maybe tell you. Yeah.

[18:20] quickly one thing on that that there's an area of voice biomarkers and the voice actually holds a lot of information. So especially I think on the neurological front there are a lot of papers that are out there that are showing things like Parkinson's detection etc that this is very very effective.

[18:40] effective and detecting from the voice. So that could be a use case. And M, the last dream, M is monitoring. And what companies like Apple and all the huge companies have done is gone into monitoring straight away. So we use a smartwatch to monitor a number of parameters, but that's going to get more refined with time. So I look at AI and healthcare in these five areas.

[19:00] And what we were discussing now was improving the efficiency of doctors. And if you just stick to that for now, so clinical documentation is what you're trying to change. And there are so many other companies that are into this and big players as well, isn't it? Then why should we think of Ognito as our first choice? Yeah, I think the sort of way to look at it is, you know, a lot of these big companies are

[19:20] sort of like a big cruise ship, right? And we're more of a speedboat, right? So I think we've been able to innovate at a very fast rate. We've been able to really focus in on markets as well and really customize our solutions very well for those markets. Now say for example, in the GCC with our ambient technology, we've been able to come up with

[19:40] really the first clinical grade Arabic ambient solution. So what we did just now we can do it in Arabic as well. So we can do it in Arabic as well. With the same accuracy as with the same accuracy as well. And we've also obviously looked at other aspects of the market as well because we spend a lot of time on with human

The Growth of AI in Telemedicine and Remote Healthcare

[20:00] centered design and we believe that every market's got its own sort of nuances and stuff that you need to design for. So we take a very focused approach but we're also very quick with our innovation and being able to provide really a great service because my old business was a services based business. So we've brought

[20:20] a lot of those aspects into Cognito and we make sure that the service that we provide really is top notch. We make sure that the hand holding is needed and especially with clinicians, there's a lot of change management that needs to be done and you really do need to be a long-term partner.

[20:40] Right. It's not just a question of, you know, I'm dumping my software and running away. Yeah. So is, I mean, of course you're giving a software after sale service and you're giving the know how of how to run it in an organization. So I know that big corporate hospitals will be obviously, you know, contracting with you to have this one. But what about smaller clinics? Do you provide this service to clinics which

[21:00] 4 or 5 maybe up to 10 doctors clinics. So you can buy one single license. Okay. It's available online. You can just download. And have all the functionality? Yes. Okay. So that's available. But if you want to sort of a more sort of integrated solution into your electronic medical record,

[21:20] Then you would have to contact us and we could get that done. But we've actually integrated with over 75 clinical applications globally. So chances are EMRs already 75 EMRs. Not only EMRs but you know risk systems and risk systems as well. So you know across clinical softwares.

[21:40] been able to integrate with a bunch of them and some of the time hospitals have developed their own HIS system as well. So we're happy to sort of work with them and provide our APIs and SDKs and sort of integrate that in directly as well. So you I mean obviously in your profile you state that about 375 hospitals that you're providing. It's about yeah

[22:00] about 370 maybe customers but more than 500 hospitals. Wow okay and the kind of feedback you get from this you're learning from it and you're trying to improve your yeah I mean in our type of business the way I look at it is you know it's a software as a service so that means you have to keep improving right and we spend a lot of

[22:20] time talking to our customers and figuring out how do you actually keep improving this technology, how do you keep improving the platform, the service that we provide. So it is, once you do sign up with us, it's a constant improvement actually that you're signing up. So we always talk about the positives of AI which we just mentioned, but many of the doctors are skeptical about

[22:40] about this whole thing. And so I know doctors who are at two ends of the spectrum. One, they think it's just a fact it's going to go away. And there are others who think this thing is going to take my job away. And I always keep telling them that it's not the AI, but the person who, the skills of AI that might take your job. So what should we be afraid of?

[23:00] What risks are there associated with this AI technology? It's a great question and I think in healthcare one of the key sort of principles really is explainability. Now if you don't know what has trained a certain LLM and it's a complete black box, according to me that's an issue.

[23:20] Another issue could be hallucinations. How are we going to sometimes these technologies can hallucinate and gives something completely wild as an answer. So how do you build the guardrails in as a software to make sure that that's being picked up as well. And there can also be biases. There can be biases in

[23:40] the training data. What's going what are you feeding into the what you're feeding in now if you feed in a certain bias it's going to give you you know like an AI created in a say a white dominated population exactly you cannot extrapolate that into a multicultural you know a colored person's community. And I would also say there's also a

[24:00] what we call an automation bias. So sometimes we may become sort of too comfortable with it as something that you mentioned and then you don't do the checks and balances that are needed to make sure that quality is. Because you mentioned the word hallucination, this is something like when an AI app or

[24:20] in AI software gives you an answer but it is not accurate or they give you a completely unrelated result. Unrelated yeah or it could be related but it could be something that is not what it is supposed to. Supposed to right yeah and that is something that still needs a check. That is still something that LLMs are still it is still a big

[24:40] problem for LLM not just in healthcare but across LMs. So what is so fascinating about this whole thing is that I was looking at how AI has revolutionized diagnostics. So just taking an example of an X-ray, we now know that the accuracy of extra diagnosis with an AI is much higher than pure radiologists put together. It's a fact we know. But what

Ethical Concerns and Risks of AI in Medical Practice

[25:00] There's a study that looked at this and they looked at all the data that has been put into an AI and now the AI can tell you what race the person has from whom the X-ray has been taken, which a doctor can't. So it's giving you answers. You've never asked those questions in the first place. Same thing with ECG. ECG telling you whether your risk of diabetes

[25:20] diabetes. I mean a cardiologist will not be able to look at an ECG and tell you what is your risk of diabetes, but an AI can now. Opens up a lot of possibilities. And now we know that there are many possibilities. It's what we don't know and it's still there that's scary at times, isn't it? Oh definitely, definitely. I think you know diagnostics is

[25:40] obviously a very interesting area to work on but it's also an area that's fraught with risk. What happens when the AI gets it wrong? It's gonna have some impact, someone may die because of it which is why diagnostics need FDA approval, all these different

[26:00] approvals and stuff like that and I think our call as a company was that I think there's an example I think in the US maybe 600 billion of the 4 trillion they spend every year on diagnostics but everything else is like the whole like administration side of things right. So that's a bigger problem to fix.

[26:20] So the way I looked at that was that this is going to actually make the entire hospital system so much more efficient. So you're looking to develop a co-pilot for doctors. Yes. So clinical documentation is obviously one tool that a doctor makes. One aspect of it. But you eventually, maybe 10 years down the line, you might have few tools for doctors to use. It's almost like a doctor having super

[26:40] powers to do things quicker and more efficiently. I mean we sort of think of it like you know Jarvis in Iron Man. I am in Jarvis. Jarvis for doctors right. So you know would you have the ability to of course you know like a complete you know medical team around you. Yeah. Which is sitting in your pocket. Yeah. Right. So in the pre-podcast discussion

[27:00] people were having, you mentioned that if I'm a doctor speaking to you as a patient and I'm taking history from you, for me to be more accurate, I might miss some questions while I'm discussing that with you. So you can have a copilot or something which pops up questions that I should ask you, isn't it? Exactly. And that will improve my efficiency and my chances of having an accurate diagnosis for you.

[27:20] So I think what's happening with this type of technology is a lot of modes have been broken. Now, there's a huge area of clinical decision support that exists and you have huge players globally that have been there. But I think even those type of areas are up for disruption by

[27:40] startups and things like that and like you were saying that you can imagine that you would have these questions coming up along with the ambient system which can nudge you along and ask you that hey have you checked this or you know you've given an antibiotic you sure you want to give a you know. So imagine a mobile just speaks up and says okay by the way you haven't asked him this or asks you to

[28:00] Yeah, maybe you need to take an antacid along with that or something like that. So that's going to be fantastic, isn't it? And sometimes these are simple things, but you can tend to forget to do that. So what's your next step? So imagine you solve this problem for all doctors in the world where you say they are at least

[28:20] seven to eight minutes per consultation which is about two and a half to three hours in a day for a doctor. You can imagine that's huge. That's done dusted. What's your next step? So I think for us we're very very focused on you know trying to develop what we call sort of our intelligence stack on top of these technologies. So we see this like as a platform and what

[28:40] What are the next levels of intelligence that we can keep building on? So there's just a huge task still for doctors to be able to provide care across the world really and to make it available to most people. So I think healthcare today still is a very expensive thing.

[29:00] do you actually bring the cost of health? And equitable I think. I want to talk about that, telemedicine, which is huge now post COVID or during COVID. So we know we can speak to a doctor anywhere in the world and have a consultation. And we might be having an error, it's already happening, where we talk to an AI machine sitting across and having a diagnosis going on. Where do you think this is?

[29:20] going? How far how much would you find this is going to be and what are you planning to be involved in this kind of technology? I honestly believe that you know how do you bring the cost of health care down right? How do you get it into you know rural parts of different you know rural parts of Africa or rural parts of India? You know how do you get it to everyone? You know can you

[29:40] get the highest quality healthcare available, someone can just check their voice biomarker while they're talking, right? You can detect and get medication over to people through drones. I think there's so many ways and models that are coming up and frankly, look, we're just at the very early stages of this today.

AI in Mental Health: Voice Biomarkers and Early Detection

[30:00] Like the internet. The early 1990s of internet. Exactly. So we don't know exactly who the winners are going to be. Obviously you got a lot of big companies you're putting in billions and hundreds of billions into the field. But I'm very confident that startups are going to play a very key role in all this.

[30:20] I think you can innovate much faster. Like I said, some of the big companies are like a big cruise ship but we can be a lot more agile as well and I think that it's just a very exciting time. What about integrating or cooperating with other kind of companies, larger companies to build something bigger?

[30:40] We do cooperate with a lot of companies. We integrate with a bunch of companies. We work with telehealth platforms. We work with a lot of people because at the end of the day, healthcare requires you to integrate. It requires you to collaborate, which is why I believe that healthcare is not an industry that creates too many monopolies.

[31:00] I think it's a massive industry and there's a lot of place for players to build great solutions in. So this technology that you're talking about, say I'm just talking about clinical documentation, it keeps getting refined. So do you think at what stage are you at at the moment? Are we looking at more refinement over the next five years?

[31:20] or in the next year or two we are going to have the final product. I cannot say that it will be the final product because I think you will always have smarter ways of doing things. I will give you an example. We have done some recent research which is a published research that we have done for radiology reporting where you do not even have to

[31:40] dictate the entire report. You just give a few findings and the report will get generated. 10x faster now. So I think like that you're just going to keep having smarter and better ways of doing things. In the future it's very possible that an MRI scan will automatically be able to create the report. It just

[32:00] understands the canon. So when all this came into the limelight, the radiologists were thinking, are we going to lose our jobs? People were thinking, are radiologists going to lose their jobs? That I think is not going to happen, at least not in the next hundred years. But I believe a human touch is absolutely important, isn't it? I mean, I completely believe that, you know, like you were saying yourself, right, that people

[32:20] who use AI will probably be the ones who survive. People who don't use AI I think would find it hard. Hard at the time moving forward. So where do you see yourself in the next five years in Dubai or UA for that matter? Well we're working very hard to really provide the best solution to healthcare

[32:40] customers and new ones coming up in the region. It's always been our goal to really be number one and we're working tirelessly to make that happen. I'm sure with this coming into Dubai, there's competition as well coming in many areas and you're obviously trying to be the best possible provider for this particular kind of service.

[33:00] There was something a concept called pajama time for doctors where you finish your work you go home and you start dictating at home to save you time next day and I know many surgeons many physicians who spend two three hours of their precious home time dictating and This AI has kind of resolved that problem for them. Isn't it gives you a chance to spend time?

[33:20] with your family. I mean what more value could we get out of that? See more patients maybe. See more patients as well. So overall there's cost saving as well with this, isn't it for an organization? There's definitely a huge ROI that comes with this. So can you just split that and tell me how for me as a as a CEO of a corporate hospital how would I save money if I take your so I'll give you an example.

[33:40] One of the leading healthcare institutions in Asia is a customer. We've done a case study with them and we have been able to show that we've actually opened up 7% additional physician capacity just with 300 doctors using the technology. So that's significant.

[34:00] Like having 21 extra doctors. Exactly. Seven percent. So that's a very powerful thing and when you actually turn that into you know into into money terms right it's significant. It was showing a 21x ROI on every dollar you spend on Neato. I'm very keen on mental health because that's something I do as well.

[34:20] my team here is involved in children's mental health and with mental health we listen to stories and we have huge consultation times of 40 minutes listening talking to a patient. What we've done now is a three minute quick consultation of an upper respiratory infection but when it comes to psychology and psychiatry we're talking about 45 minutes to an hour. Correct. Has this been tried in that field? So we are starting a

[34:40] out mental health, we see it as a massive growth area for us because just from the fact that you're talking for one and a half hours potentially there's a lot of documentation that's needed. So we see it really as a tool that will help in a great way and I also believe very strongly that it can actually be the

Closing Thoughts and Future Prospects of AI in Healthcare

[35:00] areas of voice biomarkers as well will be a big game changer in mental health. Because you can imagine, in fact, we have done some research internally. We haven't taken that forward as yet, but where you can detect depression score with like 90, 92% just from the voice.

[35:20] is for different areas. It's like a screening tool for highly scalable screening. And when you apply medical transcription costs to these kind of mental health institutes, the amount just escalates to a huge number actually and having such a tool at a disposal, say for example

[35:40] For example, in my clinic, I have a few psychologists, food psychiatrists and neurologists sitting and dictating all the time and spending that much time. If you have this with us, it's going to kind of bring down the time to more than 30 to 40 percent and that can help us improve our efficiencies. But you can imagine, I'm saying in the future, you're doing that, but also the biomarkers are giving you

[36:00] So how are you approaching your team? Are you going to smaller clinics as well and kind of showing your work to them or you're just sticking with big corporates at the moment? No, we work across clinics and hospitals as well. So if someone needs to contact you or your team, is it an email you're an email across or

[36:20] We have to send an application of how does it work? Yeah, I mean if you just go on toognito.ai and you can just sort of get on there and someone will get in touch with you. And you have your after-sales service team here, isn't it? Very much so. So just wanted to know that you have a presence here and an office here as well. Very much so. So is it okay for us to give your details along with this podcast? Yes, that'd be great.

[36:40] Excellent. Okay. Great. That was a fantastic session. I think the first of its kind in scrubs and suits. Thank you so much for your time. It was a real pleasure. Thank you. Thank you so much. Thank you. Thank you.

[37:00] you