Welcome to Real Talk on AI and I care with your host, Dr. Scott Morris. And I'm co host, Dr. Rehan Ahmed.
In each episode, we discuss current news, adding a little AI education, and debate innovative AI tools and topics that will change the industry. Well, hey, hello everybody. Rehan, how's it going? It's going well, Scott.
How are you doing? I'm good. It's been a while getting ready for, let's call it lecture season. So life's pretty busy.
Yes. And I know we're excited about the vision expo coming up in a couple of weeks excited to see everyone there. This, this week for news of the week, we are talking about a headline that struck close to my heart. It was about an AI from a company called Open Evidence that aced, like meaning you got 100% on the USMLE, the United States medical licensing exam.
This is, this is a test that I took every med student, usually after two years of being like heads in the books to all the time. I studied for like for two months, six weeks, like full time, and it got 100%. And I'm pretty sure it didn't study full time. You know, it didn't stay as definitely not as long as me.
So a couple of milliseconds. Yeah, probably. And I probably told them those seconds to do it. So again, the company behind this is not open AI.
This is called open evidence. They trained their AI not only to pick the right answers, interestingly, but also to provide step by step reasoning mainly from the well sided, journals like new and the journal medicine or the JAMA. So it's kind of interesting that it showed its work. So again, so this is a smart buddy that's reasoning.
But there are some caveats, thankfully for us in the white coats. This, if you go into the details here, Scott, this was a text based version of the exam. So that's kind of an interesting and important detail. So the tackle image based questions like EKG's or fundus photos or CT scans or, you know, all the stuff that, you know, we use an imaging and there's a lot in medicine.
It's a most of time where you're, you're combining different things. So it's an eye care that's so visually happy. It's so visual. It's so visual.
So this is, this is a very different from a single multiple choice exam than taking care of taking care of patients and clinical. And you know, Scott, you were just mentioning a study that you found where like it was like, what did you just say like 24% pass rate for where they had to do clinical context. They did clinical context in the past rate was like 24% and the worst student was like 40 something. So it was like a completely bombed clinical care.
So, you know, as we always say, it's a great tool. It's not a replacement. Yeah. And I think what's interesting here.
I mean, this is kind of like when I saw this headline and thank you, Scott, you sent it to me. It's just, it's, it's kind of you we expected it. And the when is just getting faster and faster. I know it was like maybe 70% a while back.
And it's like 90% was like, okay, well, is it ever going to ace it? Well, here we are. It, it aced it. And it provided reasoning, which I think is super cool.
So for me, the next, I think the next bar would be when will it ace the test where it does tackle and mid space questions where it is truly multi modal. And that's definitely coming. This is again, the worst model that we've, that's been created so far. So it's, it's definitely coming.
That'll still be explainable and help with education. And it's happening faster and faster. So while we're not yet putting away our white coat, you know, putting away our white coats. This is the fact that it's able to ace a test.
It just goes to, I think, really a lot of the themes that we've been discussing, Scott, that it is augmenting us in ways that we can't even sort of imagine right now with the type of knowledge and inside that it's going to create. And help hopefully us get the humanity back in medicine. That's what I've been hoping for because the gig that gives me hope for the future. But, you know, we shall see if it starts getting that then I then I don't know what to do, but that I thought that was an interesting news bite.
What are your thoughts? No, I agree. I mean, I think taking a test is one thing, interacting in a live synthetic live non synthetic environment is a different thing. And so, you know, as we always say, it's going to be a competitive tool.
It's not going to be replacement. I mean, it's just, you know, it's going to give better. And I don't think it's going to be very long for aces. The image side of either because that's just, that's just, that's just image recognition.
I mean, and it already does that better than we already does that. That's a fun way. But that's different than interacting with the nuances of humanity. So we'll see what happens.
Yeah. I did what you brought something I almost forgot to say for all of our audience, please do not forget rayon and I and Rania Abash and Eugene Chasman will be on stage Wednesday, September 17th in Las Vegas for the AI and I care event, we call navigating the future. Where AI is going and how it's getting back to all of us. So if you haven't signed up already, please go on to AI and I care.
Commentable to sign up there. So I'm never super excited about that. I'm finishing up, you know, the deck send up. We are showing, I think, never before seen footage and some audio clips and video clips of how AI is actually being used.
So super excited to see everyone. I have a video that I have a video that I'm doing around that I think we're going to hear a pin drop in the room. Can't wait. So every one of our podcasts, when we do our real talk, we have a knowledge bike because our job, our primary job is to educate all of you who are listening about all the different subject matter that's out there in the world of AI.
This week, we decided to kind of tackle something that we mentioned a lot called telemedicine and after questions, well, what is it? And we hear my term all the time, but you know, what does it really mean? And I guess in a nutshell, we can probably summarize is it's the use of technology that's going to provide health care services remotely. That's a big subject that covered a lot of things.
So I want you to think of it kind as the doctor's office and maybe the diagnostic lab and maybe not the diagnostic test and sex station all from the comfort of your catch. And so the most common example people often think about when they think a telemedicine is kind of like a virtual doctor's point, right? So I said, it's going to get in the car and going to the clinic and sitting away here for. And it's to an hour, you're going to have a real time to a video call with your health care provider.
And this is like perfect for I think routine checkups and follow up stuff. You know, even getting prescription refills or those types of things, something where you don't have to have clinical testing and touch and feel and observation and diagnostic, you know, test. And that's how we kind of think of telemedicine right now. I think after we're done with this little piece, we're going to get into remote diagnostics.
I think we look at telemedicine in the future, but maybe very different from that even. I think it's more than just video calls. I mean, it's going to be let's call it storing forward. That's a new term I've been started to use a stored forward technology.
It's like you're going to send your medical information. Let's say it's a condition. I have a, I have a brown spot on my arm or, you know, I got a lab report that I don't understand. I'm going to do a special see their dermatologist or, you know, I have the other day by my son.
Was a little instant I had a CT scan of his face and I had sent it to my ocule plastic surgeon go, OK, this is broken. This is an oral fracture is this just a line, you know, a suture line and she goes, it's just a suture line. OK, it's 30 seconds. You know, I send to the pictures she called me back to the years what it is.
I was told about us, you know, and I think as we start getting, you know, the specials can review it and they can read whatever they want. And then send back to diagnosis or the treatment plan or what it was. It's a really cool and great way. And it's a way not just of now, but in the future.
And then the second opinion and access to an expert who might live in a different city, state or country for that matter. Then I think there's a third part of this, which is remote patient monitoring. And I mean, this is not always as a game changer, but it's here now. These are people with like chronic conditions like let's say diabetes or heart disease or I think soon go a coma where you'll have this devices that are like a split pressure cup or a continuous glucose monitor.
And then you'll be able to do those in contacts before very long or maybe when we have an implantable device inside the eye for glaucoma that can measure true. Not just diurnal pressure pressure 24, 7 every 10 second measurements and AI can feed in and read all that, you know, and then it's going to send that data in some way shape or form. And it hits a certain level and it's a danger risk or things are good or whatever. I don't know about that.
But since that healthcare data, your patient to your provider who get in like track that progress and then if something looks off, they can kind of intervene. So me, why is tell medicine important? Well, I think it, you know, it's, we're now getting the points where we can break down the barriers of geographic location. It was going to make healthcare more accessible for people in rural areas, those people with mobility issues, anyone who struggles with transportation.
I didn't think it might change the way finances work, right? Cause what is it? Something like 98.6% of people over the age of 18 have a smartphone in the United States. Well, that means 98.6% of people potentially have access to a healthcare provider that they can share information with.
And so people who are people that have a car that solves transportation issues. You know, it's going to save times, going to say money, it's going to reduce travel, wait times and, you know, I think it's going to allow people aren't well and don't have access to get access. And why I don't think it's ever going to be a replacement. I mean, I remember doing a podcast like this ran over COVID.
And so, I think the medicine was like, oh, we got to launch this right because our clinics were dead stops and the patients needed us and people like, well, this is going to be the replacement. Like it's never going to replace the in person human touch. And that's what we're just talking about with your news of the day. You know, I still think you're going to have to go for a physical exam where you have testing and you get touched and felt there's interaction all kind of stuff.
I just don't think it's going to be in the model that we use today. I think we're going to a lot of what we're going to do is tell medicine and you're going to have a once every exp amount of time, depending on the condition. We actually go in to an office. I think it's just tell medicine is a way of the future that's going to make life more convenient and efficient, personalized and all those models we had five years ago when COVID was around.
I mean, those are all we would look at those now and go, those are a track tape. So I mean, technologies move so far forward telemedicine is going to have a totally different look to it. Yeah, absolutely. Interesting.
And you know, the one thing and I think it's all subtext Scott where you you didn't say this, but you did say it. This is the AI and I care and I was an intentional that you didn't bring up AI and this and past that. Well, because I can't talk about it. Yeah, I really think that we're for the audience.
I mean, now we're going to get into remote diagnostics and how that takes into the world of telemedicine. And I think we're going to bleed AI into that because telemedicine in the past was, you know, I got to be on the other than the line. I'm not so sure that's going to be necessary. I think some telemedicine is going to be with chatbots and some telemedicine is going to be with human doctors and some is going to be and maybe that doctors.
You know, I think I'm pretty darn good about in the world to try. Why can't I help somebody in Israel or somebody in Spain or, you know, whatever. Why does it have to be somebody who lives in my hometown? That's crazy.
And I think AI is going to give us the ability to do that. So, and let's kick off. So, you know, we try to do these every week when we do these is these we call first a chast between Ray on and I've it. Let's pick a subject and talk about something really cool.
And I think you and I've been talking about it and I've had a lot of conversation the last couple of weeks about remote diagnostics. Like, I think that's a fear of many of our peers. And also at the same time, like this technology so cool it's going to help us do all this and I'm like, you have to think about the goods and the interesting challenges. I won't say bad.
It's been interesting challenges of remote diagnostics look like. And I think when we think about remote remote diagnostics, people like, yeah, but it's not really here yet. And I'm like, I don't know how many of you have a smartwatch or smartphone or a ring or soon going to be smart glasses. All of those things are remote diagnostics.
It's not going to be just clinical data like let me measure IOP and look at what are in FL are. I mean, I think we get in a remote diagnostics. I think it's not just going to be external. It's not going to be dermal.
I think it's going to be sub dermal. I think it's going to be subcutaneous. I think it's going to be parental. I mean, I think it's going to be float around our bloodstream.
I think remote diagnostics are going to take a whole new turn of the way it looks. You know, I think first of all, it fixes accessibility. You know, it's kind of like we just talked about the knowledge by this. Imagine anybody if they have a device.
I've been saying this for years. I mean, I know that I've been saying this for 10 years that someday somehow somebody's going to figure out a way to put some type of nanotech micromicro small stuff in the interior chamber and measure. I think it's going to be a major change in the interior chamber and measure IOP or maybe be able to measure ocular pulse amplitude or maybe a measure, you know, metabolites within the bloodstream of the right. And those technologies are all here now like they're we're going to see those in the next year or two.
They're out. They're just waiting for approval many cases. I think we're going to look back at glaucoma go it's 2025 and we knew absolutely nothing about glaucoma. That's what I think.
Yeah, well, you know what I really love about sort of the remote sort of smart diagnostic space. I am always attracted to sort of clever technologies that are embedded within otherwise every day normal objects. For example, there's a team out from out of MIT that developed a regular looking weighing scale that every day you come out on the bathroom, you stand and you stand on your weighing scale. That's like a normal behavior that people do.
But this is a smart remote diagnostic enabled weighing scale that will look at the blood flow of your foot to better predict the risk of getting diabetic neuropathy in your feet. That's great. All sorts of other diabetic complications in the same way like an aura ring and otherwise normal looking ring. In fact, that looks normal.
Like any other ring. Maybe it's a little bit picker that does all those things that we know it does right or glasses. And this is where you know, I kind of, you know, I wish the Ray van metas and the smart looked less. I think the better the more they made you look better as like a regular pair of glasses that they're fashion forward and otherwise look normal.
I think the better for the technology, right? Yeah, imagine a time. I mean, that's just a matter of time, Ray. They're going to figure it out.
Yeah, they're going to get there. But like it's all about. It's all about like or like a mirror that you go into the bathroom and it checks your blood pressure or your blood. I mean, I love the continuous blood glucose monitors.
But that is something that's attached to your body. Is there a way we could do it without without doing that without sort of attaching something that we just look at it. Other, other sort of other signals that we can look at. I think so.
I mean, you think about are we going to build a someday put a little, you know, I mean. What was the name of that company Elizabeth Holmes. There is a there is right. I mean, that crashed in burn.
But yeah, I mean, that that type of idea will be able to put something on our skin that can do. You know, transmit brain blood blood measurements. And I think that's just a matter of time. I think so.
And the questions like, do we even need to do that? Is there other smarter with AI like looking at big data? Right. And the contact lenses and blood glucose that's been looked at.
You know, I don't know what the current status of that is. But there are a lot of so I'm really fascinated about remote diagnostics that are intelligent that are just within our normal workflow of everyday humans that we normally do that are just smarter. And so that's why I think, you know, business a lot of this cool interesting businesses. Potentially again, be formed.
That's what gets me excited. You know, he's talking about intelligent. Let's call it intelligent design, right? I'm listening to this book right now called sapiens.
And it talks a lot about intelligent design. And what the future of humanity looks like. There's a scary topic chapter 20. If you guys ever listen to that chapter 20 will definitely open your eyes about the future.
And so, but you talked about intelligent design kind of, you know, and I look at it in the broad, I didn't really cover that in the knowledge bite about where AI is, but I mean. Let's go there, right? So let's talk about remote diagnostic testing and will AI be able to recognize patterns that, you know, maybe doctor, maybe patients are going to come to us four times a year for. A glaucoma check, right?
I mean, we can measure their IOP if you look at their blood pressure and look at their blood glucose levels and look at their metabolites coming off of their optic nerve. And the AI is telling us telling them when they need to come see us because something's changed and then beating us that data. You know, right now we can make a check pressure four times a year and maybe we even check blood pressure, you know, and. You know, I mean, when are you checking blood pressure?
Of course, they're in at three o'clock in the afternoon and the blood pressures high and the IOPs low. But are we really checking, you know, hypotensive issues at 4 a.m. when the IOP is traditionally at its higher. We don't do that.
I mean, nobody does that. So I think when we could start measuring everything 24 seven and maybe it corresponds to sleep patterns, right? Maybe we're not. You know, gosh, I at 4 a.m.
is when I have my most restless sleep and or and my blood pressure is at its lowest because I'm restless and I'm anxious and you know, I don't know. I don't know what those answers are, but I think in the next five years. We're going to have answers about all of these diseases that we couldn't even conceptualize today as we just sit and talk about the future. I think that AI and I'll say not so much gender to be I know and I think that you and I was the day around as it.
We've so went down with talking about AI. We've so went down the path that everything is gender to be I. So your AI is only a piece of the whole big spectrum of AI. And I think everybody's going, well, I really like Gemini or Chappie T and how's it going to help me?
You're forgetting that 80 90% of AI has nothing to do with gender to the I and L. L. It's something else right? It's machine learning.
It's all the time stuff. And I think that this is where we're going to see kind of a move or shift back from gender to the I into true. Let's call old school AI if you call it that. You know, maybe we're too early in the spectrum to call it old school, but old school AI that's machine learning.
It's going to be able to look at these diagnostic technologies and do this automated say hey, there's patterns here. There's patterns in the diabetic right now, but there's patterns in this and you know, I mean, that's going to help us triage or maybe the I's triaging it maybe our staff triaging it maybe we're triaging it remotely and saying, OK, this is great. This is all good information. Well, I think I need to see in six months.
If the data changes will move this up. And I think about how that's going to change our flow in our in our in our clinics, right? If now all of a sudden I could sit and have a five minute conversation like this, like you and I are having right now with the patient instead of having them come in my clinic and slow it down. Is that change what my workflow is and guess how much staff I needed to have this conversation, by the way, zero right.
And so what does that look like? I mean, and will machine learning build these, you know, I envision. I envision a day and I want to throw this I want to throw this up for the vision expo meeting. Actually, I envision a day where we just have a panel and it's giving us all the predictive analytics as we're talking.
And I think in 10 years I said that when we talked about remote I wear, I think in 10 years we're going to see there's providers just going to give us the predictive analytics as we're talking to the patient. But I do think before that happens is just like you're on a zoom meeting, there'll be a predictive analytics saying, hey, this is what's going on with this patient. Here's what we predict. Here's what we suggest based on the other 200,000 patients that have the same similarity across the United States across the world.
I think now remote care telemedicine everyone to look at it. That's a whole different aspect of how we do medicine than what we do now. A lot of unknowns, because as you're talking, I sort of think, but why are we needed to, you know, like what? Why are we still need to communicate, you know, I said, but the audience, you guys are listening to us, but we actually do this over zoom.
So we can read each other's faces a lot of times and go, okay, did we go in too far. You know, but I think that when you're sitting there talking to another human being, there are so many, you know, what is 87% of all languages body language and 13% what said. We're a long way off from AI reading body language and all the other increases that mean it's going to happen someday, but it's going to be a while. I think they're still a part of humanity that it takes for us to communicate messages back and forth and just looking at somebody going, you can tell if they get it or they don't get it, you know, do we need to explain it or explain it.
But having all the tools, the machine learning predictive analytics and, you know, diagnostics and comparative evidence based medicine and even what we can in the future value based medicine sitting on a profile in front of us as we're talking. How much better providers would be be if we just had more information about that human being with this particular condition and how they compare against hundreds of thousands of people just like them all in the world. Hopefully a lot better. I think that is the hope and the goal and that's what we look to provide to our listeners and readers.
That in fact, AI will be an augmented tool that that we can use to elevate our care for our patients. So like Scott said, we look forward to seeing you Wednesday morning Las Vegas September 17th correct. So that's right. 9 o'clock in the morning and we will see you there.
Check us out in AI and I care dot com and look forward to chatting with you next week. You've been listening to real talk an AI and I care your weekly podcast to keep you informed about AI technologies revolutionizing I care.