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. So welcome to today's knowledge of AI, from AI and I care. So today we're going to be talking about big data. So at first glance, the term big data is kind of self-recreational in a way.
It's kind of obvious. You think, well, what's a big data? It's just a lot of data. A lot of data sets, and you think, well, something like the VA or EAK start from the hospital.
It seems pretty straightforward. That's what big data is. And indeed, that's more or less right. Big data are vast data sets.
I typically use to spot trends, build prediction models, all the stuff we sort of talked about in the previous episodes. But as we dive deeper into the application of AI care, there's a lot more nuance beneath the surface of the term big data. So yeah, big data obviously involves massive volumes of information. This includes all the stuff we do in AI care, the retinoscans, the OCTs, visual field, the chimitry, smaller sublime, building-osky, medical records, the wearable type.
These are all data sets. And they can do stuff like predictive analytics. And like algorithms, it's got last couple of weeks ago, we talked about what algorithms are. So these are capable of identifying patients that risk for the class, the case of diabetes, renoub, the or gual hormone.
Earlier, then traditional methods. And even more so recently, we're talking about you can big data identify things that are not obvious. Like cardiovascular risks for all time risk. We talked about that.
But we want to sort of take a step back and talk about big data and sort of what it means more broadly. So really big data is a couple of ways. So one way to think about the benefits of big data and the role it plays is considering real world evidence or some of our RWE, person's clinical trials. That do a lot of consulting and farm ones.
So this is going to come up. And it's becoming much more important, this idea of real world evidence. So clinical trials, as you know, it's got these are very carefully controlled conditions. And the patients or participants have very strict inclusion and exclusion criteria.
If you're just off a little bit by age or took the wrong medication beforehand and not perfect, you will be excluded. But real world evidence, the benefit is that it captures diverse populations and genuine clinical scenarios. So the patients who are not exactly perfect, who don't read the textbooks. And you have treatment, conformment and everyday practice.
And inside that control trials, for example, in ophthalmology we do retinal injections for macular generation or deadbed or not B. And we've learned the real world evidence that a treat and extend protocol, giving patients an injection but gradually lengthen the time between injections can be beneficial. That's been, that was initially obtained through the real world evidence. Some of the doctors just did and we found that it worked.
The second thing is super interesting with regards to big data. It's in precision eye care. And this is where we're not treating to the average. Right.
What are the average patients doing in clinical trial, but personalized to the individual's unique profile. So we detailed data from millions of patient counters. We can now scale their fees to the exact cases, make genetic makeup, lifestyle history, really increasing. And I got to get a shot up and in place to the recent AI innovators talk.
I did with Rania Dr. Rania about from basketball, and she talks about this really interesting idea of digital point that sort of fits in nicely. Imagine the future. And so an additional twin is where you can, it actually comes in supposed to have to walk home or a retinal disease.
And they asked, well, should I be on this drug or that drop and it's like, well, my drop, well, you can just run a simulation based on big data that applied to that individual to say, well, let's just run the simulation and simulation. It turns out that, you know, this medication, you know, this varies version of a cross the blind and it's better than that version of lots of that. That is where the power of big data is. So the other areas where it really plays role is data quality and bias.
So we've all of this phrase garbage in garbage dump. So with which things data, the idea is that the sheer hungry hints of this that the gigantic nature, how enormous that it. It will generalize over flaws of inconsistent data and create incomplete records or biased samples. Right.
So we're we're big data, but the huge role is giving data tons of data from everywhere and another one with the one I had with Dr. Stewart who runs it works a lot with the Ormits, which is a international eye care organization that likes a lot of data from populations that are really harder reach. I think, you know, we're one or a Bangladesh getting data from from so many areas really can supplement larger data sets throughout the world. Suppose the data sets from North America, you're there to complimenting those data sets with hard to read data sets creates elements of larger, more generalizable data sets.
And that's one of the benefits of big data. Another point I sort of want to be about big data, which also give me more expanded on how we think about it. Right. So when we talk about big data, we sort of immediately think about each are, you know, radiographic or fun, this photos, but really another great area.
A big data that in my mind has not been explored to the degree that it's going to be in the future is consumer behavior data or what are patients to when they're not in clinic, which is. Well, we know most of the time they're not in the most time there are outside of us observing that right. So these were rules stuff like the or ring which I have or some sort of smart. And I think that the other thing that's going to change, Rayhan is that, you know, as we start moving into world of us all having our own AI agents, the data that's going to come out of those you're right, how that's going to impact it.
Yeah, you know, is massive. And I do want to come on. I don't play devil's advocate on this a little bit because the challenge we have with big data is I agree with you as if you have enough data. You can somewhat gloss over some of the inadequacies in in the data or the bias in the data, but the reality that's still the biggest challenge of big data is where does that data come from.
Because you can't take it from an EHR because patients didn't give you the access to do that right there's that that's not a hippo issue or hippoconsent form you have to get get access to all of that data and acceptance from all of the patients to use that data. So I still look at it and go I think big data is huge, but I'm just not sure where we're going to get the big data from. You know, I mean, how do we teach large language models kind of same thing. Where does the data come from and I think until we have a universal platform that can actually track all that.
I think big data is still going to be a little bit of a challenge. That's exactly right and you sort of you read my mind the rest of the future because the last point was the challenge and that's exactly right. You said it very well. People are getting are smart enough about their data as they should, but similar companies, institutions and government.
So, you know, one of the best source of big data is that of course a UK biobank, which is a great area, a great amount of data. But you know, think about companies that have tons of data when they're not going to be, you know, maybe in the past they were, you know, really giving it to researchers and thinking about it, but you know, people have thought are thinking much more carefully about their data and patients are as they should. Data is the currency of the future data is a new oil, you know, and so there are a lot of ethical considerations for privacy regulations like it, but. And so you make it at some point is the days of freely available big data sets.
I think those are what I'm. And so that is definitely a challenge and I think there are a lot of companies who are thinking about how to create. Inventment, platform systems or data sharing platforms systems and thinking about how to check privacy and security and that's a very. Active area of research and activity and sort of in the commercial sector, but yeah, that is a that is a huge point and you know data we're talking about here we're not talking about clinical data or as I was mentioning we're coming insurance claim data environmental data social media interactions and so you know it is connecting all that to fun.
There's a lot of potential areas of benefit, but also abuse and so really smart how we think about that. Rayon, I think what we talk about we always do our fireside chats you know and I as we were kind of prepping this I posed an interesting question and it came to me today because I was talking to one of my patients and he said he goes because he was giving me a hard time because I start late on Tuesday Wednesdays we don't start till noon, but we do go till like eight o'clock. So it's a little different hour and he's like what do you do all morning as well. I work on other things I trained my student doctors he goes well what do you think student doctors are going to do in the future and it made me think about the question of what are the skill sets of the doctor of the future when we put AI whether it be in the form of big data like we just talked about or whether it be in the form of no longer will you be have to be the smartest person in the room to be the provider because the reality is all of that data big data.
Big data clinical data research data no longer will you have to keep it in your head you won't have to remember every different type of disease and every different type of side effect of a drug. The reality is is that all of that data will not be something you prompt and go search it will actually come to you when necessary as necessary. So it made me think as I'm talking to him well what are the skill sets of the future I mean maybe true you know healthcare providers are still going to need to understand the basics of how healthcare delivery works and some of the basics of knowledge but I also think that maybe the skill sets of the future aren't necessarily natural ability to remember intelligence but more how we look at how we can do that. How we connect you know I have this phrase above my next time you guys are watching the innovators podcast you'll see a little phrase on my bookshelf that says the ability to connect is greater than or intellect and I think that you know as I sit there and I talked to my student docs I'm like I really think that the one thing AI isn't going to do probably in my lifetime and I can't speak for you Ray on but I don't see him replacing doctors I see as I've said many times before AI is going to augment our medical disease.
Our medical decisions but I do think that the skill sets that our healthcare providers the future are going to need are a little different and it might be more about you know things like do we know how to do change management and change management can be do we know how to interact with AI inside of an office setting or through a telehealth portal. I think that's an interesting skill set we're going to need I also think we're going to train people about behavior modification right instead of being dictating as a provider what that patient should do I think in the future we might more have to figure out how to help that patient change their behavior on a 24 7 365 when they're not in the office that's the way we become better providers in the future not necessarily just what we know. Yeah yeah totally agree you know and this something I think about you know I have three young kids it's right with this whole AI revolution I think what are the skills and this is maybe beyond what are the skills that are going to be important to you and I you know. It applies beyond but definitely to healthcare I think one of the skills being generalize it right so I don't know as you said like do we need to have master diagnosticians anymore I'm not I'm not sure that we we will like it's going to be you represented that to you but you're going to need judgment.
I agree. I agree. I think you can connect and have where chair side manner best side manner are going to be very very important so I think I think those are the good like I told my kids I'm yeah I'm trying to inculcate and those are skills that I think are important. Well and you know that brings up an interesting point we're in the tiktok you know kind of world and communication skills have definitely faltered in the last really I think since covid communication skills have went downhill not improved because you know we don't have to talk to people we can text or we can email or we can you know.
I think that's a challenge. I mean well before we go into the technical parts of what you think you know I mean what are your thoughts on the communication like how do we how do we improve that I will agents build a help us with that is going to be we're going to teach a class on. So we're going to do a lot of communication will that be just as important as electron optics or pharmaceuticals yeah maybe so it would be how I run it have an AI to communicate like I mean that's that's the weird part or you know you're your AI agent talking to the patient the AI agent and they come up with a plan together because you know we're all learning this together this is a giant experiment on society at large and so our patients are also learning to. But I definitely think when you're communicating especially with regards to a life changing diagnosis or a vision changing diagnosis you know just the other day I had a young gentleman is 20 with a very severe retinal condition and you know that he started crying in the exam chair and these are things that we hopefully we can send more time with the medical students on how I'm just doing it.
So I think maybe it's not that we should spend less time on the diet but you know hopefully we'll be able to have more appropriately spend our time so that we've neglected in the past I think so I'm hopeful for that. I agree and I think you know I call it translational services because I also think that it's going to be up to us as providers to help translate or communicate what some of these AI driven insights and how precision medicine works we still got a decode the doctor to talk right and we got to be able to translate that into common English and maybe the agents are going to do that better than we are in the future maybe but I think we're going to go through a transitional period. Where we as providers have to translate what those precision medicine AI driven insights really are and we have to do that effectively to patients so they're not afraid of what AI or augmented intelligence is providing but instead learn to embrace it and go AI plus my doctor their intelligence there in the lack their experiences that makes a better healthcare team and I think that that's one of the greatest challenges I see over the next decade that I know ran you and I've talked with that's kind of why we do these podcasts right is we want to help providers and anybody else who listens to this kind of embrace this technology and go we are better together and that's not just with AI I mean whether you're ophthalmology or optical whether you're AI embracing or whether AI fearing the reality is we all got to come together to make this change and make this work. And I think that that we're entering a very different era of instead of it being knowledge centric maybe that's not the right word maybe instead of it being knowledge dictated from provider to patient we're going to move into a very much more patient centric kind of era that I hope AI is going to facilitate for sure but I there's challenges without it out.
Oh yeah I mean I believe our patients are going to have access to very similar AI tools that we have access to and it says that they don't get their retinal photographs and they'll do a lot of the AI analyses themselves and it'll be much more. So we're like they have a they're bringing it with their AI they're bringing it a deep mongolous and so the judgment is still coming from the clinician and the augmented intelligence coming from the from the provider and so I think that's where I think that's where I'm a lot of hope not you mentioned I think to one area I think technically you know I did at the last year's the next world where I talked about their public companies but one life. There not just some of the lowest rates all over the world can that really really sure about it currently and that was January following the group has to play some very hard to the public. but I still think there's a definite place for addition to improve, continue to improve their technical skills.
And so I think that will still be, for the foreseeable future, I don't, there will always be circumstances or resource constraints that are you don't have robots available that you're gonna learn. And that gives me some pause because everyone's training up, you know, very robot doing everything. Where will trainees be? Develop those specialized skillsets here.
That, you know, we're eventually, you know, many. So I think that's gonna be an important part in the future as well. I agree, but I think, you know, also, I, you know, we talk about this, you and I both have our own companies that we work with. And, you know, I think that there's gonna be still some core sets of technology, some core skills that providers the future gonna need.
They're gonna need to understand how do AI algorithms work, right? I mean, I've written now hundreds of thousands of pages of AI algorithms over the last four or five years. And, you know, understanding how an algorithm works only has helped me as a provider, be a better provider and understand how does my algorithm work in my head versus how is it gonna work from an AI perspective and where's the mesh between those two? I think that healthcare providers of the future are gonna have to understand like, hey, here's how these algorithms are built.
And here's the pluses and minuses of algorithmic decision making versus kind of, we've talked about neural networks like three dimensional processing, you know, and what does that look like? I think the docs of the future are gonna have to have a little basic understanding of that. I think they're also gonna have to, we were talking, you were talking earlier in your knowledge by about big data. Well, I think it's gonna become more important in the future for providers to understand and be able to interpret and kind of analyze the results of these data sets.
They might not be analyzing the data within the data sets, but I think it's gonna be important for them to interpret and analyze the results of those data sets instead of just saying it's gonna get fed to us from big data and a simple line of code. I don't think it's gonna be that way, at least not in the near future. I think that we're gonna have to understand how to interpret these data sets. And not looking at a study, you know, I always think about studies as having this conversational, my student docs today.
You know, lots of the things that have been done over in our career, Rayhan, have done in studies that have 30, 50, 100, 300 people. Well, the reality is the data sets, the future might be looking at hundreds of thousands of patients across a variety of different genotype phenotypes. And how do we interpret that data? And I think that's a skill set that some, it's gonna be instead of memory, you're gonna have to understand how to think through this interpretation of these data sets.
Yeah, 100% agree. 100% agree. I don't know, I think it's gonna be fascinating. I think that, you know, we're still gonna have to understand how to do critical thinking.
We're still gonna have to understand basic decision-making. And I'm not so sure those are always taught directly in schools. I think those are a byproduct of our education. And maybe instead of being a byproduct, they should be the primary focus, critical thinking, decision-making, change management, behavior modification.
I almost think we should have a course that's called those things, right? These are the skill sets of the future, or future skill sets or something like that. There might be just as important or more important than pharmacy pharmacology 101. Yeah, I think the truth is, it's core things that, you know, I was a philosophy maker.
So the core thing, the suck-proud of method of question and answer, that's never gonna go out of sight. So I think that's definitely gonna be the part of the future. And what you know, it's really interesting, one of the best sellers from an edge, you know, I was reading an article in terms of what educators are buying. Do you remember those plain blue note pads?
That's not the thing, I think, but just college, like college rule of paper. Yes. Yeah, that's definitely going to be the best seller now, because everyone's using their AI laptops and now they have teachers are demanding that they just write with a number two pencil. So there is definitely an experiment, a work in progress.
I do not, I do not envy teachers right now. Oh, not at all. I'm not interested in this. But, you know, me, I'm super excited at this massive piece of intelligence that's right next to me.
I just hope it doesn't turn out to be our collective range to much. Yeah, I agree. I think it's about working. And we said this from the very beginning.
I think in every podcast we said it, it started with when we did the meeting last year, is I think the future is about augmenting our intelligence with AI, not replacing our intelligence with AI. All right, well, as usual, Rayon, I enjoy our conversations. I love our probing into the future. And I think, you know, I kind of wish 20 years from now, we're having this call and going, huh, we were kind of right.
And we were way wrong. You know, it'll be somewhere in between there. All right, everybody. Well, thanks for listening in.
Stay tuned. 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.