Is Healthcare Ready for Digital Twins?
Loading...
What if the most powerful diagnostic tool wasn't a scan or a lab result — but a continuously learning model of your patient, built from their own data?
In this episode, Dr. Junaid Kalia sits down with co-host Dr. Harvey Castro, this time in the guest seat, to unpack one of the most transformative concepts in modern medicine: the healthcare digital twin. They trace how the concept evolved from NASA simulations and factory floors into clinical reality, breaking it down clearly for those new to the idea while grounding it in real applications — predictive analytics, personalized pharmacogenomics, and Harvey's personal digital twin, which he built and uses today. They also don't shy away from the harder questions: data privacy, implementation costs, the global adoption gap, and the tension between rapid innovation and regulation that can barely keep pace.
Whether you're a clinician thinking about what's next for patient care, or a healthcare enthusiast trying to make sense of where medicine is heading, this episode gives you a grounded, honest look at a technology that's closer than you think.
"The digital twin provides us with way more than just predictive analytics — but in a way that is continuous."- Junaid Kalia, MD
Loading...
What You’ll Discover
[00:00] Intro: What Are Healthcare Digital Twins?
[04:28] The Human + AI Concept: Harvey's Personal Digital Twin
[06:00] Top 3 Applications for Digital Twin Transforming Patient Care
[07:05] Data Privacy and Cybersecurity Risks
[08:29] Why Culture Is the Real Rate-Limiting Step
[10:22] Implementation Challenges in the US vs. Other Countries
[12:14] Are Tokens Reliably Used in Healthcare?
[15:06] Genetics, Epigenomics & the Future of Point-of-Care Treatment
[16:30] The Future We're Building: A Final Word from Two Doctors
What You’ll Discover
🖇️ Ask Harvey a healthcare AI question –– https://harvey.deepfoundai.com/
Transcript
Junaid Kalia, MD:
Good morning everyone. Welcome to Signals and Symptom podcast. Today we're going to have a special episode and I'm too excited about it. My good friend Harvey Castro, which is generally a co host, we're going to treat him as a guest today. And the reason behind it is that I really want to learn this very important topic which is essentially healthcare digital twins.
How you go from concept to a clinical reality. And given his experience internationally, we have a better outcome or at least see physically where things are going to be in the next year or so as compared to here in US where there's a lot of issues with data silos etc, and more importantly how important it is.
So we're going to work on this as usual.
So without further ado we're gonna go ahead and let our Harvey Castro teach us about number one digital twin. And then we're gonna talk about digital twin healthcare.
It is again a harder concept. So for our audience, let's just first step back and remind them what the word digital twin means. And then we're going to talk about healthcare and then we're going to talk about its applications.
Harvey Castro, MD, MBA:
Yeah. So big picture, looking at this way, literally it's kind of interesting we were following NASA, watching what the government's doing.
Well, back in the day, NASA was having trouble in space and they didn't know how to fix the problem. And they literally had to create a mirror image of what was going on in space so that they could stimulate and see how to fix it.
Well, that concept was eventually what became a digital twin. And, and it's actually all around us and we don't really realize. A couple of years ago I spoke at the Consumer Electronics show and that was the main theme was the digital twin. I spoke with Siemens and basically what they were doing is creating factory digital twins.
So if you wanted to start a factory, they would go through and be like, okay, here's what that factory looks like, here's the process, here's the workflow and all the components of how that machine would be working and talking and friction points. That way when they would develop it, they knew what they were developing down to the nanosecond, down to the centimeter, the whole thing.
Now let's extrapolate that down to healthcare. What does that mean in healthcare? Well, again, different flavors, depending on the use cases. And so personally I've been really, really excited because we'll go through it.
Junaid Kalia, MD:
Before you go into Harvey, one more thing.
So digital twin, I understand. Right. So for example, I'm designing a car, but then I design it on a computer simulation as a digital twin, put it in a tunnel, which is, you know, also digital.
But have you had any other examples of non mechanical digital twins outside of healthcare industry?
Outside of healthcare industry? What else have I seen? Obviously we've seen airplanes, right? That's a simple one. We've seen how they would act, with the air resistance and the different models.
So, yeah, one of the, that's where I wanted to. Before you talk about health care, I want people to understand that how it is so different. Different. We talk about digital twins from designing this tool to this table, to everything simulation wise. And not to mention the whole AI robotics, which is we're doing gives you very small data set now that then it generates simulated data set which is then converted into a digital twin and then converted into new AI, robotics.
And honestly, I mean it has massive, massive global implications because, production pipelines are not going to look like what it looks like today.
It's insane. I mean, there's a whole protectionism globally going on. And one of the reasons is that now healthcare is different. And I just want to classify that because when you're doing digital twins outside of healthcare, you're mainly talking about mechanical things. And then in healthcare.
Go ahead, Harvey. Sorry.
Harvey Castro, MD, MBA:
No, no. So in healthcare, like you said, it's, it's, it's an interesting being. And, and I know this sounds a little philosophical, but it really is all about that human plus AI concept that I always talk about.
I started playing around with my own personal digital twin. And as an ER doctor, I thought, what does that even mean?
I thought, well, I see people dying in front of me and I thought if I could leverage AI to give me predictive analytics to tell me before the incident happens, then I Have something and that's literally my personal version of the digital twin that I'm creating more predictive analytics, more looking at data.
How can I change it? As opposed to. If Junaid made it, he would look at it a little differently and he'd be skewed to his personality.
Junaid Kalia, MD:
Before we actually in predictive analytics only was pure statistical, but now we have hybrid mechanics and ML surrogates, that he just described beautifully in simple terms as I'm doing it stupidly.
But anyways, what I'm saying is that the idea is that, that the digital twin provides us with way more just predictive analytics, but in a way that is continuous.
If you have access to other digital twins as well, we can do predictive analytics at massive level as well. Basically doing contraflexure as well. So what excites you most about digital twin technology in healthcare?
That's number one. What is the, top three applications that you think are going to change the way we function? Yeah, that's a good one. I'm really excited and it's really more of a category I'm really.
And I guess it breaks it down, but from a big category. I'm excited about predictive analytics. I'm really excited about how can I create a digital twin that gives me predictive analytics of, let's just say Harvey, and then taking it to the next level.
I'm really. And these are all things that are coming, or some of it's already here is now that I know what Harvey looks like, then if, let's say you and I are taking the same medication, how can we change that medicine to personalize it for me, for my genetics, for my weight, for my age, for my way of metabolizing things.
That to me is going to be the difference because now we're getting personalized care because of that digital twin. All right, let's see. I got, we got a question. How do we address the key challenge like data privacy and implementation costs. Awesome question, Kathy.
Obviously I'm blessed in that I'm leveraging these tools. And so I literally have my own computer, that is locked down, that it's in its own cloud that is in this house secured and I'm able to put all my genetics, my wearables, my electronic records.
And so as far as the privacy, it's not on ChatGPT server, it's not on Junaid's server, it's on my own personal server and I have it locked. And then as far as implementation costs, yes, I'm an n of 1 and so yes, I'm spending the money but for me this is the way I see it.
I am fine tuning it, fixing it, creating it and seeing how can I leverage this technology for everybody so that those implementation costs goes down so that the privacy is there.
This is one of many reasons why I'm working for Phantom Space because of what you just said. My vision is can we create a digital twin and then leverage the sky, leverage the data centers so it processes that information, comes back to me and let's say you on your phone and then through blockchain technology is going back and forth, leveraging that information.
That to me is gold and it will basically address all of those things, lower the cost and keep the privacy. Absolutely perfect. Would be. Understand. So now going back to this. So I see digital twin in healthcare is number one, what you just described for personal wellness.
Personal wellness means that, you're not only using your Apple Watch, sleep device, etc, going with, essentially with your health data and then giving you predictive analytics for you know, and also recommendations hopefully to improve your health overall, which is very important.
So we have level one, level two, and level three in terms of healthcare.
Now tell me where, what is the current status of level one? Do you think it is with announcements from Perplexity Claude? Are we, are we there or are we not there? And then the second thing is, of course, when do we get to level two? And then in your situation, in your mind, when will we get to level three?
Harvey Castro, MD, MBA:
You see, to me, it's quite interesting. Our rate limiting step is ourselves, right? Our imagination is there.
Our technology is exponential, but the culture is not. I talk about this all the time and I do want to mention it. So there's something called Martec's Law. Here's the skinny. AI, algorithmic. It's going like a hockey stick straight up. You and I, where we work for, it's more of a logarithmic change and slow.
The bigger the gap becomes, the more disruption. The reason I'm going over this is because this digital twin question answers that question. It's because of that. If I am in Singapore, if I'm in the, UAE, where they're looking at digital twins, where they're creating the framework to make this successful, then implementation is going to happen way quicker.
But I do have one quick prediction. We will in my opinion, see other countries doing this, create FOMO here in the United States, call our congressman, call our President and say hey, why don't we have it?
And we're going to see a huge shift all of a sudden here in the state. So you know, time wise I don't know but I see that in the next few years we're going to be moving quickly. I totally agree with you. I understand data is the biggest issue. There's the highest risk of cybersecurity and this what is other challenges that you see in us versus the rest of the world that we need to address today.
Going back to that analogy of the foundation, we need to make sure that we have the infrastructure, that we have the data centers, that we have the energy that we can drive the cost down because if the cost is too high, it's going to happen for the masses.
Right. That's one of my my, my, I don't know, my soapboxes right now is how can we lower healthcare costs? Leveraging AI to help the future. That is it. Can we lower the cost of the infrastructure? Can we have the data centers?
Can we have more open llama models? Open models where we're able to leverage them so we're not having to pay as much in tokens. Can we create more on device processing like the iPhone, that are coming out and more and more of those things that we can leverage so we can bring the cost down.
Those were the things that I would, I would be pushing my, my congressman. And by the way buddy, in I don't know, let's say eight years, my goal is to run for one term in Congress. So just one time.
Junaid Kalia, MD:
And then after that we are going to be very excited, you're going to have full support and then we're going to basically do this for you.
And I hope that we have forward looking congressmen like you that understand this because it is also as, as you always suggest, cultural shift. And that comes from leadership explaining to everyone that how important it is.
And I'm just going to be honest with you. I mean the way that I'm looking at LLMs in the future is what you exactly described. It is like a meter that I'm paying for electricity is going to be tokens, period.
And we talked about the cyber security risk, we talked about the levels of autonomy that's gonna, it's gonna create. what are your predictions that a, a person can reliably use tokens for healthcare.
And do you think that this surge and everything is going to improve overall, Harvey? Or do we really need to push in this regard?
Harvey Castro, MD, MBA:
No, I, I, I do see things going to the edge, meaning using all these AI, tokens.
Eventually a lot of the uses is going to be on your phone. So you won't need to pay OpenAI certain things because it'll be running here. Taking that analogy, then I see that same concept happening where there'll be many AIs happening in the office where you're, the doctor's not having to pay OpenAI.
Junaid Kalia, MD:
But do you really think, given that you are using it personally as well, do you really think that AI in healthcare will bring true healthcare outcomes that are going to be better than what?
Harvey Castro, MD:
If we could do one single thing, do one predictive analytic, we can change the future.
Let me give you a quick example. Unfortunately, my dad died, in his early 50s of a heart attack. Long story short, here's the point. He had a heart attack. No warning signs. Good analytics. But here's the skinny.
My mom is alive and she had cancer. She had ultrasound, herself found a nodule. Long story short, she was able to get chemo, radiation. She's alive. What is the difference? My mom, let's call it, had predictive analytics because she caught it early.
My dad didn't. My dad's no longer here. My mom, 20 years after that concert, cancer, she's still here. So with that principle, can we leverage AI to lower the cost and to save lives?
Junaid Kalia, MD:
And lastly, I mean, Harvey, let's just assume that you are Elon Musk right now and you have access to technology that's going to Release in three years and maybe not release because of whatever that reason may be.
But anyways, what I'm saying is that, you really think that these level of technologies already exist or are they, are they too far in the future to provide this level three personalized medicine, in the, you know, at all?
So the number one question, do we have this stuff?
Yeah. Do they have the next iPhone or the next two, three generations? Yeah, they have them. Is it release? No, unfortunately. Is it economics? Is it technology? Is it culture? What's holding it back? So I do think the answer is the information is there, the technology there.
Harvey Castro, MD, MBA:
If I was Elon Musk, It's a catch 22, right? Because some things, if this sounds weird, we don't were not ready when chat GPT release, what they released. Was the world ready for all the stuff they were thinking and creating and had in their vaults?
No. Oh. The reason it was successful is because it was slow. It was to go back to my earlier statement with Martec's Law, it was logarithmic, but they were just pushing it, pushing it because they knew if they showed this, it's not going to happen. A derivative of that statement is startups call me.
They're like, hey, Harvey, how do I start this at the hospital I'm like, dude, do not show them this widget. Do not show them this other widget. Because this they can accept. Culturally, if you show them this, you're going to scare them away. You're not going to get that deal. And so with that analogy, it's the same.
Junaid Kalia, MD:
So now just step back for the audience army and make them explain where, where this genetics and epigenetics comes in.
Insane, right? I mean, I'm going to take it one step further. We are now officially ordering genetic tests, especially for kids when we're starting medications for seizures.
Because we now need to know that there's 1 in 10,000 chance that, you know, something bad can happen. But more importantly, we now know that some medications work in better genetics and epigenetic situation. We're already doing it. It is part and parcel of today. But what I'm saying is that, that in future you're going to have one test, you're going to have access to this.
You will be able to share through your digital twin to your, you know, provider. And then, you know, at the, at the, at the, you know, at the point of care, you will be able to adjust all of this.
Harvey Castro, MD, MBA:
And then I do want to highlight the digital, twin that I put in the chat. Basically, basically, oh, I do want to show that. And we'll pull it up when you're done.
But basically I created a rag model. What it all, all it is is my talks, my books, whatever interviews I've had. And I put it into a database and then I put that website that I just linked. And we can show it. There it is, harvey.defile AI.com but the skinny is you can call me.
I actually created a hotline. So if you scroll there, there's actual phone number there. And then, you can text me on your phone or you can go on the website, ask me a question. It's a little geeky. And then it'll process. It's using my clone voice. Then it'll talk to you from my clone voice, and then it'll generate a video.
And then it takes a bit, five minutes later than that image that you see there becomes animated. And I start moving my head, I start speaking. It's my clone voice and it's answering the question you just had. Again, this isn't for this fluid conversation going back and forth, but I just wanted to show proof of concept and I Was able to Amazon all by myself.
Junaid Kalia, MD:
This is truly amazing. And think about it. You can actually put this into, you know, all your data connected with this. And then Harvey is going to tell you every morning, which is odd. I don't want Harvey in my phone. But it is what it is. The world we're going to live in. That's what he's trying to show you.
That this is how the future lives. And then again, I am again most the mechanism of AI generated virtual control arms. And think about it, that not only for me personally, I can actually do virtual one or two, even digital twins, stack them onto one another and then basically create this.
Which again, I just, the last question of the thing, how are we gonna, you know, regulate this? Because this is, I mean, I'm gonna be honest with you. This is literally confusing to me, Harvey. I mean, are we gonna be able to regulate this?
Harvey Castro, MD, MBA:
No.
It's gonna be tough, right? Because people like me will be able to. And you guys, eventually the technology and the how to will get so simple that you'll be able to do it. Your. We'll be able to have our own digital twins.
But that's the part that the regulation needs to be considered. Because if someone is creating these healthcare digital twin and does it wrong and hurts themselves or it's going to be an issue.
Junaid Kalia, MD:
So Harvey, we talked about regulation. This is, honestly the last question of the day.
How your theme about culture is going to impact regulation moving forward. So how do we, you know, have this matrix of regulation self autonomy increase? It's insane. The level of increased speed of development come together, especially now bring it to the international, you know, comparison as well.
Harvey Castro, MD, MBA:
Because on one side we want, like you said, you have all these models that's helping humanity and we want those things. But at the same time, like any tool, if it's used wrong, you know, then people may die from it.
So how do we find that balance? And unfortunately, sometimes it's regulation to bring that balance to the table. You and I have the heart of a giving, heart of a doctor. And so we want to see these things come true. But unfortunately there's other players out there that you know, will use some of this technology for bad.
This is really, really powerful. And so that kind of worries me. And then the other part of this is, let's be frank. If my digital twin says, yeah, Harvey's gonna, he's gonna bite the bullet here, he's gonna die here in about five years, my life Insurance, my health insurance.
You know, who's to say that all of a sudden I don't start getting denied or certain things start changing because of my digital twin? I don't know that this is going on because it's being manipulated against me. So things to think about. That's where we. We need the regulation to protect us.
Junaid Kalia, MD:
And it just. It is really getting, truly, I mean, astonishingly, different. I mean, it is, I don't know, it is a little scary at times. But it is what do we think is going to be the best of times as well, if you make good choices as he suggested, and then implement it in our cyber secure, privacy preserving way.
© 2025 Signal & Symptoms. All rights reserved.