Our path to an AI doctor - Neil Naik, MD

0:00 We know in some instances what brings contentment joy and sometimes happiness Nourishing and building social relationships taking care of our minds taking care of our bodies But we still find ourselves chasing money status power and myself as well My question to you is would you be open to an AI solution? That tracks your your social activities your social media your your email and even controls it, tracks your biometrics, if it will lead to more contentment in your life? Oh yeah, 100%. Right? So happiness is the spikes that we feel, the endorphins that we feel all the time. And I think if we overwhelm our bodies with constant endorphins, it becomes the new norm. And then we're always seeking for the next pleasure, the next day. I think that's what we see. We get the money, it goes for our bank accounts, we get that spike of endorphins, or I want more of that. That felt great. How do I do this? Whereas contentment is more so the state of I'm happy with what I have. I'm not saying that you shouldn't be striving for more. You should always be striving for more. For me, it's education. Knowledge is power. And so that gives me my contentment and gives me this happiness. to know that I'm learning something new. And so the two go hand in hand, but they're not the same. And so absolutely,

1:32 I'd want an app that does contentment. I don't want it to, I feel like sometimes social media is there to give a rise out of somebody, right? And so they can give you the middle ground, and then occasionally they'll send you something that they don't know whether or not you're gonna suddenly be like, That's my guy, that's my gym bro, that's my passion project, or I don't like this politically decisive, I don't want to do this anymore. And it's the reactions. You can go back to Facebook all the way from back when we were in high school, where you could like something. And then people wanted the ability to go beyond a like. Let's get a heart, let's get a thumbs down. At one point I remember there being a thumbs down, I'm not sure if you remember that. Or a dislike. And then at one point it disappeared again. I don't know what it is anymore because I don't use Facebook anymore. I use Facebook for marketplace. I need to sell stuff to my office.

2:28 But that's the rise we want. And then at one point we wanted more people to have likes on our posts. So then we'll say more controversial things. And then Facebook will take that and run it through everyone. And so I think that's where we're starting to get into this endorphin rush of I need to have this. I need to have this dopamine rush. the addiction that we're looking for. Contentment is the removal of that addiction. I'm happy with where I'm at. I'm happy with my own thoughts. I think that's the third part about this, of this AI app, is does it make you think, and does it make you think independently? Or are you the type of person, I don't want to say you, Rashad, I don't want to say anyone, are you the type of person who listens to somebody else, and if it sounds smart, You'll just say yes Because your own thoughts are a scary place to be How did you get interested in archaeology?

3:28 Oh That's from that's from job that I my mom jokes. I just cuz I like gold. I like shiny things and King Tut's tomb was shiny and gold My mom used to teach us that Indian culture is very similar to the African culture and Egyptians, the Krishites, stuff like that. And, you know, as a kid, your parents were always right. Now, I just watched a National Geographic channel yesterday, a video on the pyramids of Sudan. And that's the ancient Krishite empire. I'm reading into that as well. Yeah, yeah, all that stuff actually is starting to line up. And, you know, we also have studies that now show that Verbal history is sometimes as good as written history in cultures. And so I think from a young age, my mom just kind of imbued that into me in terms of explore the past, have an open mind. Just because some experts said it doesn't mean that it's factual. It doesn't make them a god. It doesn't mean that they're 100 % right. Challenge the status quo. If we don't challenge the status quo, we'd still be using the Yellow Pages. because that was the absolute best thing in the world back in the day. So challenge the status quo, I'd always ask questions. And if you prove it to be incorrect, great, at least you proved it.

4:59 But to outright say no, because you don't like it, that I can't tolerate. So that's how I got into archeology, which is challenging the status quo to my mom's... She's upset now because I decided to take that to a whole different level and started challenging what is Hinduism and reading Vedism and reading the texts of the Vedic texts. I've read the Bhagavad Gita. I've actually gone off and read the Torah, the Bible, the Koran, any religious text I could. I think it's a really fascinating way of looking at how cultures change over time and how people perceive and how cultures perceive that change and what does... What is the different culture of society? And so those books are also, in a way, a historical archaeological text, because it's the interpretation of reality into a book. And then you also want to line that with what the groups have with Strabo's books and what Alexandria was. And that's kind of how it all got together. As I said, it's all clear pathways. And I think that's representative of what our We are a mixed group of really smart animals that figured out how to create a civilization out of this today. Yeah. You know, when I did residency, I was out in BC and it was really my first exposure to indigenous culture. And then they had us go to powwows. And it was very

6:31 interesting. And one thing that really, you know, stuck with me as a almost a fact that we don't respect Indigenous culture is that it's an oratory culture. It's a verbal culture. Contracts are verbal. They're not written. And we force their leaders to adopt to our framework with written contracts. He said something very interesting. And I think the world would be a much better place if our verbal word was a contract, if we did not need written contracts. Now you need proof of that someone said something and I think that that's accurate but I don't think it needs to be pen and paper or you know keyboard and screen. What are your thoughts on verbal versus written contracts and how did you learn that a verbal history or storytelling and stories are you know I'm curious because I would think they're less accurate than written history but from what it sounds like you're saying that they're They're as accurate. Yeah, it depends on how much emphasis on accuracy is placed, right? So we all see the studies on court cases and forensic cases where people will change what they remember.

7:52 And it's not maliciously, it's just the brain can fabulous and changes things. And so what they recollect even a day after a week later, whatever, is different. What I think the difference here though is, what do I remember unintentionally? I was walking down the street, I saw a car crash. What do I remember? And usually nowadays you and I, when you write down our statement right away, that reinforces what we remember. So it's less likely to change over time. If you don't have a written language, what is the next best?

8:32 thing you can do to reinforce. You share that knowledge with somebody else. You share it immediately. You can either be doing it through song. So now you have a rhythmic pattern that reinforces a memory and grammates your brain. That if it doesn't rhyme or change with the tone, you know you missed the part. So now you're actually creating a checksum into the actual memory itself. Then you can translate that to the next level. So horizontally. through the generations and vertically through the generations. And that's what allows it to pass down from one generation to the next. And I'd argue that even if it's a written contract, it's not as accurate because it depends on the context of when that document was written and the environment around it. And also written from a physical nature point of view, written contracts actually don't persist.

9:27 Let's think of the Dead Sea Scrolls, right? We need to use x -ray vision now to unroll them. They don't persist. We know they're really old. They're really ancient, but they're not as good. But oral traditions actually have that. So I think they both have their pros and they both have their cons. I think where the oral piece and the written piece differed, where I do like a written contract is I don't have time to put my EMR end user license agreement to a song. I could be singing for hours on end about why I can't. how we're going to get sued and how TELUS is not liable for anything. And so I think there's a value to that. But I think when it comes to things that are supposed to make us feel proud about our culture, to warn us about things that are important, to tell us about important milestones in our species' existence, song has a way of hitting different parts of our brain and making us feel differently. It's a reason why we sing the national anthem. A recent old day... back in school, because it brought us together and made us feel proud about being Canadian, that I know we're true north strong and free.

10:37 And I will always be with me. I don't need to write that down. And I know that everyone else around me is the same. So I think that's the value of the written word versus the verbal word. And the culture of writing and stuff like that, I'm sorry, for the Inuit piece, you're right. We don't value that, unfortunately. We should. And we now know, and we have been saying this for ages. Well, they've been saying for ages that their written word has been accurate. We're confirming that with archaeological data. And it's because somebody challenged us as well.

11:17 We said, oh yeah, 10 ,000 years ago, they came across the Bering Strait through Beringia. And that's how it happened. There was a study that came out two weeks ago that showed, hey, actually that may have been 20 ,000 years ago. Hey, by the way, there may have been three migration patterns in North America. And so, you know, Vikings at one point is a more modern one, but there's some talks about how, you know, other groups from Polynesia, they migrated over at a different stage. And so... We need to stop thinking about things in a two -dimensional state of, this happened, therefore this mustn't happen. And start thinking from an open mind of, yeah, we know one of the components. What else? How else? And we suddenly put an attachment to ourselves saying, let's imagine I'm an archaeologist. I'm not saying this is what archaeologists do, but I have heard stories and I have read books and papers and people who defend their theory so because they're tied to it. They spent 20 years doing this one research project. And for somebody to come along and question it, that's a whole challenge. Let's take Einstein. He does not agree with quantum physics. It goes against a lot of these fundamental values he had. And he actually spent the last little while trying to fight it.

12:40 We know quantum mechanics works. We can't explain it. And maybe in some future state, we will prove again that quantum mechanics doesn't work. But we really need to be able to say, just fighting it, let's work with it. Yeah, sorry. Yeah, I think when it comes to things like quantum mechanics, we have an inherent desire to focus on causation and the why. But there is this school of thought, and maybe I'm the only person in the school of thought, is all we can know is correlation. I don't think we'll ever figure out causation. and we would advance our technologies and our society much faster if we simply focus on correlation.

13:23 There's need for trying to figure out why, I think, as slowing us down considerably. Do you have any thoughts on that, on my belief there? And this belief kind of was supported by this investment firm called Renaissance Technologies. um the most successful investment firm ever, 66 percent annual returns for 30 years, more than three times the Warren Buffett returns. Their whole philosophy is we will look at the correlation and they're their mathematicians they made an algorithm to do this between public equities and if you know the price of rice in Brazil goes up 10%. That means the price of Alibaba in China will rise 5 % tomorrow for six hours.

14:10 And they're they're this precise, they're on the average whole time for stocks is two days. And then they invest, they remove the human element, because they say we worry about the why and the why is not what's important. Because the only thing that matters is the what and the correlations. and causations are impossible to know. There's too many infinite confounding variables and factors in our life. But we're still stuck on this, bringing us back to the VC world and the venture capital world. The biggest question is, why are the founders doing this? And my thought is maybe that is not the biggest question. The biggest question is, will the founders stick around or are they capable of doing this? And knowing why may or may not lead to answering those those questions which actually matter which is a what you know will this be a unicorn exit so what are your thoughts on correlation versus causation to be over indexed for causation and are we hurting ourselves in doing so so interesting um i think you're in your own school of thought uh i'm happy being in my own school of thought so that's amazing frankly to me I disagree with that. I think the why in the sense that VCs actually not even, they do have a why. Their why is make money, have shareholders, dividends. If correlation gets them to that end goal, then Wyoming can't go for it, right? I'm not saying that's wrong.

15:42 They obviously have proven that's right. I think the beauty of this is can we translate that to a different sector afterwards, right? So if that rice price in Brazil goes up and Alibaba's whatever goes up for five minutes. then knowing the why underneath allows us to predict more. And I do believe that at some point there is no such thing as infinite number of variables, it's a finite number of variables. We seek to achieve that, figure out the digital twin, and then figure out with that digital twin how does the universe work. And it goes back to another philosophical thing, if we know all the variables of the universe.

16:25 and we can put it into a digital twin, then we just directly just create a copy of our own universe digitally. Which then leads to the idea of are we just all part of one large... Can we know all the variables in the universe? Can we know why? Or is it impossible to know why the price of olive oil goes up because the price of rice goes up? Is that a futile question? No, I think we can. I think By saying we can't means that we've given up. And that goes back to our question of the status quo and stands still to go backwards. Humans as a species moves forward. If curiosity drives us that way. And correlation can show us the why. Correlation gives us the patterns, but we need to dive deep. I have decided in my work and not my personal life to focus on the correlation path. And I think let's have this conversation in five years and see how we've uncovered the truth and knowledge in the universe and what, let's share notes then. But let's go on these. Correlation is important. Correlation gives us, you know, of all the things in the universe, of all the variables, what are the things that are related to each other? Yeah, and my thought is that we can know we're not smart enough information we're limited by five

17:57 senses I think and as we are now in humans in 2025 We can know that that's where I've kind of Agree in 2025 humans can know we're not that smart. I think that's a I do believe in a hybridized world of AI Where is all correlation? Like AI doesn't understand why there's no meaning or significance. It's essentially, as you said, it's all probabilities. It's what is the next character? And it doesn't even know what a word is. It says what is the next half a P or character that comes after this? And the way it defines character is very different. So if we're going the AI path, then it's in my... you know, defense, I guess. This is an interesting debate. It would be all correlation and no causation, because AI doesn't care. There's no caring or not caring, right? It's just all probabilities. Do we, do you think we're anchoring AI then by making it, or forcing it to explain things to us? But if we had to let AI go, it'll create its own language, which we know it can do. and have a conversation with itself, which we know it can't do, or with another AI, it can run experiments to determine why. It can't do that, right? It's a cold start

19:27 AI, we have our machine learning AI that can then change its own base model for learning and then re -evaluate a situation, exactly what we do on a day -to -day basis. And so we're forcing it to adhere to what we want. from a language point of view of, yeah, P is the next letter. But to be fair, we did the same thing, right? I before E, except after C, and a whole bunch of other weird and wonderful rules that we had to memorize back in school, right? That's literally what we're doing for AI that we had to do. And we just programmed it into our brains. That's what allows us to communicate with each other.

20:07 So we're teaching AI how to communicate. We're also starting to teach it how to experiment. And the more we get into it, experiments, the more extrapolation it's going to be able to do. I can tell you now, when I was a six -year -old kid, extrapolation was pretty bad. I could hallucinate. I could do make -believe things and imagine that my Power Rangers toys were fighting the Transformers and all that other wonderful stuff. And that's imagination. That's hallucination. That's what we, Google's had their AI data. turns off, right? And we actually have a paper that came out last year that showed that giving AIs and their models of break in between will actually improve their accuracy and AIs do fatigue. Interesting. Yeah, yeah, so AIs fatigue is if you ask a complex question, just like you and I, it will get decision fatigue. And the answers get worse and worse.

21:06 You need to give it a break. And so It is coming closer and closer to that singularity of being smarter than us. That's okay. Because I think it's smartness and our smartness are two different smartnesses. And together that we actually can become this newer species as a whole. I kind of equated to the old schools. Are you a Star Trek fan? No. Okay. Original series? Spock? Her? Kirk is the emotional individual who's the lead. Spock is the logical person who's the science dude. Sorry, chunky fans who listen to this.

21:53 I said science dude. But the two together are better than them individually. Us and AI are better together. And if we don't go along with it, sure. But that just means that you now have a logical being. that is going to be pure logic, which is bad. And we've seen that going bad. Or you're going to have an emotional being, us, that's going to go along on their own. And that's bad too. We see that in the world today, that logic and emotions, sorry, emotions in general lead to war and hate and suffering. The two together lead to something beautiful. I think there's a full circle moment here where I'm maybe realizing all causation is infinite correlation loops.

22:40 And if you're trying to figure out the why, you're just trying to break down two things that may not be linked. And you're breaking on the path, and that's just predicting probabilities along the way. I don't know if I explained that well, but I'm kind of realizing causation just might be, or the why might just be the what repeated over and over again until you get to the why. So why does a car go this speed while you break it down? you know why do the wheels turn this fast what's happening in the engine and it's just all probabilities of what's um but you know i think there's different paths to getting to understanding the universe um i'll pick correlation you'll pick causation and we'll see who gets closer um and you know you make a good point there like the y is actually just a high probability of a correlation your correlation would be 99 or it could be one percent and so you have 99 percent you got your y That's one percent. There's a correlation. Everything has a correlation being on this planet is a correlation. But doesn't mean that has causation. I can see what your logic is coming from. Yeah. We are both physicians in the innovation space. So, you know, maybe let's talk about health care a little bit. Do you have when do you have to go, by the way?

24:03 I got to go around. I got enough. What is the path to the AI clinical position? Should physicians be scared? And then where does the opportunity lie for physicians here? Yeah, and you know, that's a great segue there, right? So you're either the clinician who doesn't use AI, definitely do that. But there's a great book called Deep Medicine by Eric Topol. I'm not sure if you've read it. 2019, if I'm not mistaken. And in the book, He talks about how, you know, in that particular year, we as a species have identified 10 ,000 diseases in humans. I am not memorized all 10 ,000. I'm pretty sure none of the doctors have been memorized all 10 ,000. And so that's where the value of AI is. It's protecting the patients in the sense of making sure that all the rare and weird conditions are identified and ruled out. It's ensuring that communication happens in a proper way, in a standardized way. There's good enough evidence now that shows that doctors, in the beginning of the day, spend more time with their patients than the doctors at the end of the day. If a patient's performance goes over time, then you're trying to get home. You need food. You need the meeting to go to. So you're more likely to give the antibiotics at the end of the day than at the beginning of the day.

25:31 And that's not hard, right? Antibiotics harm people. communication improves people's lives, not just from a social perspective, but also from an understanding and compliance perspective as well. And so when we talk about clinicians being afraid of AI, we're essentially saying clinicians are afraid of finding a solution that, one, we may not understand, but two, we're actually afraid that could be our replacement. I'd say it's never gonna be our replacement. It could be. It could be our replacement, but I don't think that's gonna happen anytime soon. it's more likely to compliment us. And in fact, probably make it, most likely it will just make your jumps really easy. So one thing that we do right now in our office is we do our AI scribe. I've been doing it for about a year now. It builds my notes.

26:20 And I've had three patients say, I like that note I wrote. Can you just email that to me so I can review it at home? Absolutely. Right? One patient hand out on some hip flexor stretches. great, go online, do a deep research on gen GPT, say I need some hip flexors, definitely check it, make sure it's correct, and then send it off to the patient, right? So instead of me having to spend the time with the patient, which I'd love to do, don't get me wrong, but there's only so much of my time that I have for my patients, for my patient who might have an amputation that I need to modify those stretches, they're getting a better quality of care. And overall, patients and community are healthier. So actually by not using AI, you're doing a disservice to your community.

27:09 I think we need to start showing that ROI is back to the patient and to the clinicians. And we're seeing that, right? So AI scribes have gone from last year of 5 % market penetration. Last I heard in the Waterloo region, they said about 20%. So we've gone past the cancel. That's still low. I would think it would be more. There's a lot of physicians I know, and maybe I just have a different collection of friends here. Yeah, we have innovative friends, right? So within our circle, yeah, absolutely. They all use it. But it's the circle that we're not a part of. So if you think of our innovation curve, our innovators make up that first 5 -ish percent.

27:52 We have the early adopters that push us to about that 9 % to 11%. then jump that chasm. We've jumped that chasm now. We're at the 20 percent. We're going to see a rapid uptake right now until we hit about 50 percent. It'll slow down eventually and we will never reach the last five percent. That's okay. They're either going to be retiring soon or they're going to miss the boat. When do you think the completely autonomous AI physician is happening Brian Marie Care. Ah, interesting. Cool. So I just came to talk about this in Sarnia, actually, two days ago. We need to stop thinking about this as us as doing one thing. We do, in the sense that we ask questions, we find a pattern, diagnose this probability and correlation. That's what you said. We test that theory with our physical exam, our labs imaging, things we do, and then we go ahead and treat. And then we do a self -reflection saying, cool, did we do it well?

29:08 Did we not? What did we miss? All that other stuff. We're getting better and better over the years. But we think it's this large continent. In fact, it's actually, we are a combination of different machine learning algorithms. We have our structured algorithm. We have our shortness of breath algorithm. We have our whatever else algorithm. Algorithms feed into other algorithms. And what we're probably gonna start seeing is an AI for cough, an AI for heart failure or shortness of breath, an AI for anemia. And what it's gonna start doing, it's gonna start branching off and taking off little chunks of our job. And it's a good thing and a bad thing. One, it helps to, scalable, which means more people have access to care. It democratizes health care. One of my whys, going back to what our car was fishing about an hour and 15 minutes ago, my why in life is universal health care for everyone no matter what. And if that means I lose my job, OK, because I need to make sure that I do no harm. And if the AI can give you the care you need at home, so you don't have to take time off work as a single parent or a person who has to work multiple jobs, that you have to decide between paying for food or rent and seeing me in my office, by all means, go use it at home. That's the most important thing, not my job. And so it will start

30:38 picking up chunks off of my job. And we're going to have to, we're going to be forced to as clinicians to reinvent what our value proposition is. So we're not going to lose our jobs, but our value proposition is going to change. My value is no longer going to be like that prescription. And that's the debate we're having right now in Ontario where we're seeing other professionals doing things that we used to do, whether it's going to the pharmacy and getting a prescription for some basic things that it's seeing a nurse practitioner, whether it's seeing a midwife, whatever it is. And so we're seeing that as hey, is this something I used to do and now somebody else can do it and they're not going to say professionalism?

31:24 We're probably going to see AI jump into that circle as well. We're probably going to start seeing it actually tackling some of these conditions that were delegated or that other professions have. I'd argue that the first sign that AI is coming after or is going to change our value prop is when it actually starts sticking on those 17 conditions that pharmacists can give medications for. patients perceive this and know it's an algorithm that pharmacists have. Patients also are untrusted computers more than humans. We did a study in my office with the Canadian Mental Health Association on the use of tablets to screen for mental health conditions, depression, anxiety.

32:10 And in this feedback form, the... The reason why people were more truthful on their tablets than talking to me was because, and I don't make you feel bad about the first anxiety, I recognize that as a condition, my patients know that, we have a great relationship, is that the perception that humans judge and the perception that computers are unbiased, we know AI is biased based on training data, but we're getting better at recognizing that. And so if our AI is unbiased, repeatable, can go beyond just the certain conditions, and I can still get the same service, well, I'm going to use the AI over by pharmacists to get my eye drops for my allergies, or to get my treatment for urinary tract infection. And that's probably going to scale up really quickly. So we're going to start seeing the pharmacist losing that volume.

33:06 targeting us as well as clinicians. We're going to start seeing that volume drop as well. And we're going to start seeing just segments of our profession go away. What the AI is probably going to do that's going to be better than us is when they start taking all those different conditions, you build an orchestra of it. And that's your injected AI. And you correlate that runny nose and the itchy eyes together. You and I do that all the time. Allergies. But can the AI do that? The moment it can start correlating between them, that's when we start seeing our jobs being replaced completely. And we've got to change our value problem again. If I'm a founder building in the spaces, say that I'm building an AI diagnosis and prescription model algorithm for UTIs for bladder infections. What piece of advice do you have for me? And what do you what data do I need to present to the CPSO? For them to say, okay, I feel comfortable giving this AI a license to prescribe macrobate or UTIs because that you know once and this is already happening in the state there's a bill proposed and there are medical boards and They're open to looking at solutions and giving them a license to prescribe a medication. What are your thoughts here? What piece of advice do you have for founders? And then how should I be positioning myself to the

34:39 college or the medical boards in the US for them to say, OK, I have to give this person a license because of the amount of lives you'll save or the decreased wait time or the and you know, the physicians will fight this. I think we both know that. What advice do you have for this founder? Because we hold technology to a higher standard than we hold humans, right? So let's take my stethoscope. Me and my stethoscope are about 60 % accurate, right? Maybe 50 % to me. But we're not 100%. But the perception is that we're 100%.

35:19 Or maybe, or it's not even 100%. Patients know that we're not perfect. And so maybe the perceptions are we're 70 or 80 % right. Right, I know. I'll just say one thing. This is very interesting because for pneumonia, the best and I did a presentation on physical exam finding, sensitivity, specificities and residency. The most accurate physical exam findings in pneumonia is a gopherny and tactile fremitas. It's not the stethoscope, but we still use stethoscopes. That's very interesting. We use that even though we know that that is not the most accurate. That's all right. I actually use, I use a Gopinia anticoagulant on all my patients. It works, man. It works. So I volunteer in Tanzania. I don't have x -rays there, right?

36:06 I'm doing that, right? I'm identifying potential tuberculosis and bacterial pneumonia for an island of like, you know, this population of Kishore -Warelu is like 400 ,000 people on that island. And they have one x -ray machine. Right. And I'm pretty sure the dose of radiation would get off that thing is higher than the CT scan. So I feel like I'm really reluctant to my patients off that because it's not their fault, right? It's just older technology and we work in a capitalistic world right now. For the non -physicians, could you explain what a Goffini -intactile pharmacist does? Because I realize they might not. Absolutely. No, that's a good point. Yeah. So I love this topic because it has everything to do with physics and nothing to do with medicine. So sound travels further in liquids versus airs. That's why whale sounds can travel kilometers through the ocean, hundreds of kilometers, and sound travels through walls and solids better than liquids. So your lungs are filled with air, my apologies, and when you get the pneumonia, you get swelling.

37:17 Just think how if you hit your elbow and get swelling, you get fluid buildup. So your lungs decay with the bacterial infection, your body starts to fight it, it swells up, you get liquid in it, and that's why it gets harder and harder to breathe because the air and oxygen can't get through. And so the sound of the letter B in air changes to bay. And so when you're putting the census scope on it, you get the person to whisper B. And it's not a full whisper, it's just a soft tone. There's a sweet spot. You should hear bay, and anyone can do this at home. Roll up a piece of paper, put one side against your ear, put the other side against somebody else's back, and tell them to whisper B. And if they're not sick, you should hear bay, bay, bay, bay, bay. And it's different from listening on a stethoscope, where you go, we call it the Christmas tree model, where you listen one side to the other, you compare in contrast, compare in contrast. With Agaphani, you go down. And so you do two long, sweeping lines of address. as we go bay bay bay bay bay but if you hear B then usually that means sounds are transmitted way better which means there's a consolidation now that consolidation just tells us the consolidation liquid is in there that could be heart failure that could be pneumonia that could be cancer and but the accuracy on cancer is really poor so i wouldn't i wouldn't Um, the second one is, uh, vibrations, right?

38:49 The sounds of vibrations. The tactile, Fremidus is exactly what it sounds like. Tactile, you put your hands on the chest, and it can be through clothing, it has to be on the chest. And Fremidus, I don't know what the Latin etymology of that one is, but, uh, it is, you put your hands on it, and you can actually feel the vibrations on the sides of your hands. You gotta make sure you put it between the ribs, because the bones get in the way of the conduction. You can actually feel the vibrations, uh, through your hands. So two really cool and very accurate ways of diagnosing pneumonia and the old school way back before we even had x -rays. That's how the house call or how the doctor used to diagnose people. But yeah. Yeah, that's what I can put in the technical part is what was the question that we were trying to answer before? The question, you know, it was no, I apologize because I asked you. It was what advice do you have for founders who are trying to get a medical license for a solution to focus on the example I use with cystitis, but any one of these conditions? So imagine if I fell asleep in a lecture at a med school, and I'm really, really bad at the subject. The most damage I could ever do to a society is just the, you know, 1 ,500 patients that are in my practice.

40:09 I don't have that high of a reach. Therefore, the damage to our society is a lot less. You give that to an AI and scalable, even if it makes a 1 % mistake, the impact and damage it can do will be far more than what I could do. And so the need to fix hallucinations is incredibly important. And I know there was a debate at one point saying, no, it's better than humans. No, no, no. When it comes to human health, and we do see ourselves as the dominant species on this planet, losing a person's life is bad, and we need to make sure we don't do that. And there's levels, right? Like, is this as bad as, hey, you bruised a finger, but what if that bruise isn't a bruise and it's actually a lymphoma?

41:01 I just went to a dermatology conference. You saw lymphoma? Shows up as cutaneous. I can take a manifestation of that, right? That is something that we get wrong all the time as clinicians. And I hope AI can do it better than us. But what if that gets it wrong? What if the training data's off? And all that other stuff. So number one thing, you will never get a clinician who wants an autonomous on -zone AI. So number two is how do you bring the human into a loop? Because we need to make sure that that helps. And number three is how do you make sure that human does become complacent and just says, cool, the AI is right all the time. I'm going to just start saying, yes, yes, yes, yes, yes. What we actually need to do is. is get it to start throwing mistakes at us, that the AI know is a mistake, that keeps us on edge. So those are your big three. And knowing that when you submit for that application to Health Canada, that you save that. It's a novel idea, it's a new way of doing it, but actually it helps to build complacency, or prevent complacency.

41:59 So number two then is, how do you, even if you've been in the loop, will ensure that humans don't feel like they're being replaced, and getting that, and then the... that actually leads to a medical directive. And so I don't actually look at this in terms of medical directives. The probability of a bill being passed, while it's being put in place for the Congress in the United States, I don't think it's passed there yet. And in Canada, it hasn't been put in place for it. We're still fighting interoperability. We don't have the equivalent of the Cures Act up here in Canada, unfortunately. We should, and it's kind of sad that we don't at this stage. But we need that. And so we got to start with medical directives. And that's something I've already looked into. And we're looking into doing medical directives for just putting a lot of algorithms in my clinic at this stage. And so by doing so, that's the first step. And if we show the value of that, and the scalability of that, and the life saved with that, just like what you said there, Rashad, then we can start saying, hey, what's the next step? How do we become autonomous? How do we help with the unattached patients? How do we ensure that we don't burn out our clinicians who are taking on more unattached patients? If you have clinicians in the loop, you have to pay them. And I think most clinicians are something that is completely automated. They would do it for $25 a pop if it takes a couple of minutes. But then you're increasing the cost of this solution, although still better than in person, because that's $37. Yeah, so I can

43:29 see. I think that's a model that can scale. The other way of looking at this is if you're in a capitation model is the more patients you roster, the more money you can make as well. And so if you do a blood and capitation model, the more availability you have, the more likely you are, the more ability you have to actually attach more patients to. What tech are you currently using in clinical practice? Oh, yeah, that's a great question. the simple things, right? Like it's the stuff that has existed in every other industry except medicine for the longest time.

44:13 We truly are one of the slowest industries to take on new tech. Voice over IT, right? Having a tree chain in our voice system allows us to actually help to navigate and coordinate patients. We use it, it's not technology. We actually use a call center for our phone calls for our patients. Makes it super easy in terms of... make sure that we have constant accessibility to our schedule for that patient, and for them to feel like they're getting cared for. Now, as anything, and I make this joke in regards to our emergency room wait times, if you make it more available, if you drop that call, the wait time on your call system, the volume will increase naturally. Because if you know it's easy to get a phone call in, you'll naturally just get more phone calls. It's an orborous, the state that needs itself. And so we're now looking at, okay, can we introduce, like almost like a Google Duo before it even hits our call center that then actually can answer basic questions. And we're trying to figure out how to solve this hallucination problem as well.

45:30 Where does AI fit into this? Is it just? is an entire tree algorithm, and we just make it that easy, and then hit zero, you'll get a human, but you put across your answer through another way, or is there a different way? So we're experimenting with different AI algorithms on creating a phone tree that's adaptable. We're looking at, so that's option number one, and that's super cool. That one actually exists. There's a company called, there's a company I just spoke with, they're called Strelo. Strelo AI, they're out of Hamilton. And they actually have an AI secretary for your phone. So a couple of clinics actually already use them. Highly recommend trying it out. We're still doing our impact on that and trying to figure that out. The next one that we're using here is online appointment booking. So in order to reduce the volume of appointments online and for folks like you and I where we're just busy and I'm not going to stand on the phone saying, well, what about May 6th? What about May 7th?

46:30 What about May 8th? No, I just want to look at your calendar and book it. And so that one was interesting because when we launched that in the Kishore -Warelu area, there were different clinicians with different hesitancies with using all -independent booking. No, I'm not a massage therapist. I'm not a manicurist. You need to be triaged. And I'm like, actually, to be fair, I don't need to triage my diabetic visits. I don't need to triage my hypertension visits. What I need to triage is if it's urgent. And right now, those urgent patients can't get through my phone lines. So I need to get all my non -urgent phone lines so that my urgent patients can get onto the phones. And in order to do that, there's a concept of where can you find the easiest solution as quickly as possible. It's not a concept, it's just common sense. If you log on to your doctor's portal, and the next available appointment's a month out, and you get angry. You're like, how could this doctor do this to me? Blah, blah, blah. And you call, and the secretary says, yeah, I can book you in in three days. Why would that patient ever go back to your online appointment booking platform if they knew they had an option for a non -urgent appointment to get it done in three days? Versus, if I threw 90 % of my appointments online, and you go online, and you're offered an appointment in two weeks, and you call into my office and you're offered two weeks,

48:00 nothing changes. Then one, that patient's gonna be angry because it's two weeks, because expectations are perceived for that, expectations will always be a challenge. But two, next time they need another appointment, they know that what they're gonna get online is gonna be exactly what they get on the phone. The only exception there is that we throw this on our online appointment booking that acute issues, not emergency, just acute issues call. But if you call expecting the same day appointment for a non -urgent thing, we're just going to give you the two week appointment, not the same day one. So that's the second piece of technology we use. Third piece of technology we use is PrescribeIt. It's a product that comes out of Canada Health InfoWave. It allows for two -way communication between pharmacists and primary care providers or clinicians in general because there's sometimes questions. Sometimes patients ask questions. Sometimes my prescriptions don't make sense or we need to make sure that, you know, the insurance companies will pay for it. And essentially, the fact is back and forth, which requires a fax to come in, my friends have to read it, identify the patient, put it into the right file for me to read it afterwards.

49:09 This just goes directly to that patient's chart and messages the most responsible clinician. And so prescribing has reduced the turnaround time for our patients getting their medications, which increases their compliance. and improves, sorry, it decreases fax failures. Faxes fail all the time. It's a pain. Third technology we use, AI Scribe. I've been doing that for about a year. I actually used AI Scribe as part of my capstone project for one of my courses in my MBA, my executive MBA program. Back then, about a year ago, I only used it like that to host visits. I did that primarily because I wanted to make sure I had eye -to -eye contact. patients.

49:56 I didn't want to have half by my heavy -inch and drop notes into a computer system. I am fortunate I can type and look at you at the same time, but it's still a divided detention. And so by having that two -way or that one -on -one session, talking to the patient, one, the visits are actually shorter now because I'm actually getting deeper and better quality visits. But number two, my notes are better. So what I need to advocate for that patient is time off. I can go back to my notes and articulate exactly why. I can provide the notes to the insurance company to tell them, yeah, this is pretty bad. This person's suffering.

50:39 We need to support this individual. And so AI describe has changed that whole game. So insurance notes now. So historically, when I first started eight years ago, insurance papers used to take 30 minutes to complete. They reduced down to about... 20 minutes with another technology I'll talk about, and now with the AI scribe, so we're looking at like five to 10 minutes. And that actually translates directly then to, these are all uninsured services, to my patients that are actually not having to pay for the right time. It's based on OMA's and physicians' hourly rates that we charge for these insurance forms. So my hourly rates that theoretically, the amount of time I spend on that form drops, and therefore my patients are paying in cash. And I don't feel as stressed. I hate doing forms. I don't feel as stressed anymore. The next technology we use is e -forms. So e -forms get sent out to the patient pre -visit. In that, they get their consent form for using an AI scribe. So they're made aware the entire time of what's happening. If they book a particular type of visit, hypertension, diabetes, IUD, weight management, things like that, they get a questionnaire that's just specifically tailored to that condition. And so then it asks them a whole bunch of questions, which then facilitates that conversation. Going back to mental health, it will do something called a PHQ -9 or GAD -7.

52:04 Those are two questionnaires that help with monitoring a person. You know this, Rashad, but like for all the listeners, it helps to monitor some of these symptoms over a period of time. A lot of times when you're in the depths of a mental health crisis, even some improvements are hard to see because we're... we are a species that self -devocates ourselves. We're harder on ourselves, harder than others are on to us. And so by doing those questionnaires, it's an objective way for you to self -evaluate and do that self -referential meditation we talked about earlier. So by doing so, we now have objective measurements that we can say, hey, in transdefinite, this person's PHQ not iron went from like a six all the way up to a 17. And now with counseling, it's going back down to a 12, so please continue to pay for that counseling because it's helping this person. And so that's really changed the game there on the key forums. Let's go, only important booking. Oh yeah, e -referral. So the next one we use, and this, this goes back to like 2018, 2017. Developed in Cambridge. Launched at Cambridge Memorial Hospital. Great innovative team down there in Cambridge. expanded Kitchener, Waterloo, and then expanded across the province, it allows us to try and send referrals to another source without using an access sheet. And the advantage of doing that was twofold. There's one disadvantage.

53:30 The disadvantage is that my secretary having to build that referral, I do, but technically I could have made it that my secretary had to do it. So that's still a possible workflow. But for me, I just select all the things I want to send to that referral. Hit a second button, it attaches it, throws it into a browser that then sends it securely to the receiving sites. I don't really have to fill out that entire referral form because it automatically pulls it out with my electronic medical record. But I sit there with my patient, I turn the screen to them, I'm like, all right, cool, do you need an MRI? Type in MRI, hit wait times, and I'm like, oh, if you want to drive to Peterborough, I can get you an MRI in 50 days. If you only want to go to Stratford, sure, that's about a 10 week wait.

54:21 But now we've increased transparency and we're both balancing the whole system. And so now people are getting, you can get wait times, you can find services that you had no idea existed. I just discovered one that does behavioral medicine with CBT in Toronto. So I sent two patients off yesterday with that. Great communication. So we're going to see how that goes. And so e -referral has changed. Oh, I forgot. E -referral has changed the way I practice because it meant that I had to get everyone's email addresses. But it also meant that I actually had to start reading my emails because if they wanted to know what their next MRI was or what their book, they actually had to go start reading their emails from Ocean.

55:02 We used Ocean for our e -referral platform here in Waterloo. And so that was really helpful because it offloaded my front staff for having to call each person and say, hey. memorize books for this particular date of time, this location, to the entrance, as much parking cost, blah, blah, blah. That's all gone. And patients are there every step. They keep me accountable. They know that I sent off the referral. When they call in saying, we haven't heard back with that referral, I'm like, yep, you know as much as I do. If it says they accepted it, but they haven't given me a date and time, I don't have a date and time for you. When you know, I know, and vice versa. we're looking at now is how do we start automating the post -visit technologies.

55:49 Thank you for that very thorough answer, Neil. There's a lot of physicians still purely clinical and they're not doing anything outside of clinical medicine. There's a big majority of them, I find, that are wanting to get involved in startups or consulting or innovation or entrepreneurship or even doing something more administrative that's non -clinical? What piece of advice do you have for them? It's a great question. I was just at a town hall for one of the organizations I'm in Pfizer for. And something stuck with me. You have to spend 95 % of your time to find the problem.

56:37 If you don't know what the problem is in your clinic, You don't know how to advise startup companies. So analyze yourself and analyze your own clinic. And it goes back to that Chinese proverb, if you're standing still, you're moving backwards. If you think your clinic's perfect, oh boy, you are so far behind. I have so much knowledge in my office, we are still not there. We're still trying to figure out what the next step is. Because that's when ideation starts. What is the problem we're trying to solve? and then what's the solution to that problem. If we start off with the solution, then we need to create the problem to then solve it. There's only ever been a handful of times when a solution is a newer, better way than the problem itself, but to be fair, the problem always existed. Nobody's ever took the time to actually find it. So one of the first things I did when I first started So I'd say about nine months in, about a year in, was I sat behind my friend's desk and told them what it was about. I'm not here to criticize you or judge you. I'm here to learn about what your role is. Because if you guys ever need to take time off or if we suddenly have something that happens, I need to know what you guys do so that I know how to do it myself. So how do you put a fax in? Where are all the steps to that center? What do you call patients for? How

58:08 long do those conversations take? I had a stopwatch. I had time to end those conversations. Again, not to criticize my front staff. They're a fantastic group of people, but more so how can I help support them? Then I also, at some point, I've done it three times now because the roles change all the time, is what does it cost you as a clinician to rehire for that role of your front staff? Because if you keep offloading all the stuff you don't want to do to your staff, they will burn out, and you will lose them. People will look for a new job. And so loyalty is important, though you're wrong. You can get loyalty, but you can't buy loyalty. We have this perception, we can't, and you can't. And so work with them, figure out solutions, co -design with them. Will this work? Will this not? No. people who don't like change. And there's groups that do like change. I think trying to find the right team is really important. There are some teams that don't like change and that's probably not the right team if you're looking to do this. But assuming that you do have the right team, yeah, co -design with them, find the solution, implement it, do it on a pilot basis. A lot of companies will give you that pilot aspect or work with a startup company and develop it. advise them, help them pivot, try and figure out something that actually solves the problem. Now if a company comes to you and says, I have the solution, you

59:39 should always ask, what was the problem that you were trying to solve? Well, and the number one thing you should say is, well, when I was a patient, I observed X, Y, and Z, and I'm like, yeah, as a patient you saw the tip of the iceberg and not all the stuff underneath it. Dive deeper, what else? Who's your expert panel? Are you that expert panel? Do they value your opinion, and if they're dismissing you, run away. Because patient experience and provider experience are two very different things, and they both go hand in hand, but one does not trump the other. So, it's really important to co -design, figure that out, and know what your office does for the demo. And then go talk to one of your friends. What do their offices do? How do they solve those things? Because an N of one is not really valuable.

1:00:26 You need to know what other offices do. Thanks for coming on, Neil. This has been great. And thanks for being open and sharing your thoughts on all the different things we covered. No problem. My pleasure. Thanks for having me. Always happy to be back.