Healing with Data, One Byte at a Time

Join us on Healing with Data as we explore "Healing with Data, One Byte at a Time" with Dr. Harvey Castro, Strategic Advisor for GPT & Healthcare.

We'll discuss AI adoption, alignment with clinical requirements, ethical considerations, and collaboration between healthcare professionals and AI developers.

Learn how AI integration, addressing health disparities, and enhancing staff AI competency are transforming healthcare. Dr. Castro shares strategies for ensuring data integrity, navigating ethics, and contributing to the AI revolution.

Tune in, subscribe, and let's harness data to heal, one byte at a time!

🎙 AI50 Podcast

📹 Want to receive our videos faster? SUBSCRIBE to our channel!

👉 Visit our AI50 website

👉 Schedule a demo

📰 Receive our weekly newsletter

👉 Follow Hanh Brown on LinkedIn

🏛 Follow AI50 Business Page

Find Harvey on LinkedIn: https://www.linkedin.com/in/harveycastromd/

Transcript

Harvey: 00:00:06
When I was in med school, I had a really famous person that came to the hospital and all these medical students were logging in to that patient's record because they wanted to see why they were there. And so being able to have that strong compliance monitoring to make sure that we're regulating who has access to that data and then where's that data going? Is it going to ChatGPT? Is it staying in our hospital servers? Is that data secure?

Hanh: 00:00:33
Hello, I'm Hanh Brown. Welcome to AI50, where data and innovation come together. We use cutting edge AI technology to create advanced language model applications. These applications are designed specifically for the people age 50 and above. Our mission is to understand and cater to their unique needs and preferences, ensuring they get the most out of modern technology. So now, picture this. You're the chief medical

Hanh: 00:01:04
officer of a busy hospital. You're working hard to provide high quality care in a changing healthcare world. You feel like there's untapped potential in data your hospital creates every day. You know that ensuring this data, Could change patient care for the better, but you're not sure where to start. Well, that's when you hear about Dr. Harvey Castro, a visionary doctor and strategic advisor for GPT and healthcare. Well, Dr. Castro has a unique background. He's practiced medicine

Hanh: 00:01:43
and led healthcare systems. He's become a leader in innovation in AI and healthcare. He's an expert in making AI technologies work with clinical needs. Dr. He creates environments where people work together. He also navigates the ethical issues of using AI in healthcare. Well, um, this has made him a popular advisor for healthcare organizations. Well, today we have the privilege of talking to Dr. Castro.

Hanh: 00:02:14
We will explore his fascinating journey. Uh, we'll gain valuable insights into how AI is changing healthcare. So join us as we learn from a true innovator. Discover how his guidance can help you unlock the full potential of your health care data. So, Dr. Castro, welcome to the show.

Harvey: 00:02:42
Hello. Thanks for having me. I appreciate it.

Hanh: 00:02:45
Yeah, absolutely. Well, we're thrilled to have you here today. To get things started, could you share an interesting fact or a story about yourself that many people might not know?

Harvey: 00:02:56
Yeah. So, I was working in an emergency room, and literally, I got called into the office, uh, I call it the principal's office, saying, Hey, doc, we need to have a sit down with you. And I was really worried. I was like, Oh my gosh, did something happen? Did I have a bad outcome with one of my patients? And they're like, no, we, we see that you're doing great. We want to see if you can move faster. And I, and I said, well,

Harvey: 00:03:20
can you pull all the stats? And so they showed me all the statistics. And I'm like, I'm actually above the average. They're like, yeah, but we've seen you at work. And when you want to work fast, you can work fast. But we want to see if we can get you to see more patients and I was really upset with them I was like, I I don't understand like am I hurting someone is something going on and they're like we want you to order less tests we want you to spend less time with your patients

Harvey: 00:03:42
and Anything that the ER doesn't can not have to do send it to the family practice doctor And I got so upset. I said to them, my goal is to be a doctor. My goal is to make sure that every patient is taken care of and I'm not doing anything wrong. And I got to the point where I called the owner of this company I was working for. I was one of the partners. And I said, here are your shares back. I went out and he looked at me like I was crazy. He's like, Harvey, Do you realize how much money you're giving away?

Harvey: 00:04:14
And I'm like, yeah, I can't work in a place that they're going to ask me to move the meat and to push people a certain way. I ethically, I cannot do this. And so I walked away, uh, fast forward, the company sold, made all this money. Uh, my business partners call me like, Hey, this could have been you. And I said, no, you know, I sleep well at night, like I'm okay. And so that's something that nobody really knows.

Hanh: 00:04:37
Yeah. Wow. That's great ethics, commit, uh, commitment and conviction to serving your, your patients. Wow. Thank you. So now, what inspired your journey from practicing medicine to becoming an estranged teacher advisor for GPT in healthcare?

Harvey: 00:04:54
You know, that's such a good question. I've always, I call myself a little bit of a nutty professor. I enjoy seeing technology, seeing something that's going on and seeing how I can fix it. And so I was literally in the emergency room coding a patient, and this was when the first iPhone had just came out. And I'm looking at the phone and I told the nurse, Hey, we need to start to strip. Let's take care of this person. They're in cardiac arrest and they're doing CPR.

Harvey: 00:05:21
And she went and got a textbook, started thumbing through. And I'm looking at her thinking, no, no, no. Like, we need to do something quick. And so I had that aha moment that I said, you know what? I'm going to create the first IV Med app in the world. And I did, and I published, and it went viral. And it was such a success where people were like, man, this is saving lives. Like, the old way was taking us a long time.

Harvey: 00:05:43
This way, it's seconds. Fast forward, I jokingly and lovingly say this thing called ChatGPT came out. And I'm playing with it, and I thought, oh my gosh. This is gonna change humanity. This is gonna change the world. And I remember being so excited. I was in this office I ran to my wife and I said, honey I'm gonna write a book on how to use ChatGPT in health care and she looked at me She's like, what is ChatGPT? And what are you talking about? And she thought I was kind of crazy She's like, honey, no

Harvey: 00:06:12
one's gonna read your book. No one knows what ChatGPT is Don't waste your time and I actually like paused and I thought you know what? You I have a conviction. I'm going to do it. So three weeks later, I was selling a book on how to use ChatGPT and health care and fast forward. It's opened up all these doors and companies calling me to give this talk.

Hanh: 00:06:36
That's awesome. Thank you for the work that you do so much needed in the health care. So along the way, how do you ensure that AI technologies is Align with clinical requirements and maintain data integrity?

Harvey: 00:06:55
That's an awesome question. Number one, especially healthcare professionals listening to this. We need you to be on the field with us in AI. How do I ensure this is we have to have healthcare professionals working with the big companies to make sure that they're solving the solution the correct way. If someone like me is not with them. They're going to fix it in the way they see it, but they don't see it the way the front lines see it. And so that's the number one. We need collaborations

Harvey: 00:07:21
with our clinicians. Number two, we need to make sure that our data is validated, that we're making sure that we're using the right data. Case in point, ChatGPT, not many people know this, but when they were training it, they were paying about a dollar an hour to people out in Africa to give it a thumbs up or thumbs down. Well, that sounds good because if ChatGPT, you ask it a medical question, the average person may not know. And so they may say, Oh yeah, that, that sounds good to me. Thumbs up.

Harvey: 00:07:51
What the people don't realize is there weren't doctors giving that thumbs up, thumbs down. And so that is an example of how maybe this technology is not fit. And so to answer your question, I'm a big proponent that we need to have our own large language model, our own GPT, but call it, I'm an ER doctor, GPT, ER, where it's an ER Doctor.

Hanh: 00:08:11
Industry. Yeah. Industry GPT. Basically your own operating system.

Harvey: 00:08:16
Yes. Yeah. And so that would be another way that way. We're complying with the standards. We're making sure that the state is correct. We're reinforcing the data the correct way to go into the weeds. We're creating like a GPT where if I ask a question like, Hey, this person has chest pain and it gives me the different options. Then a graph that shows me how did you come up to that conclusion? So then me as a doctor can look

Harvey: 00:08:37
and say, okay, now I see how the computer got that diagnosis. It put these things together. That's one way. The second thing to go into more detail would be this thing called RAG. Basically, um, imagine if I'm an ER doctor and I had my own clinical notes and then say you had a question, those went to my notes first before it went to ChatGPT and then it would explore that to answer your question. So you're getting a more accurate look at it.

Hanh: 00:09:03
Not only in terms of content, but context, right? And it's, it's timely, it's accurate and it's current. That's the kind of the whole idea of reg. So, no, that's great. I think what you mentioned are like the huge key components, uh, to reap the benefits of AI. So great points. So now in a recent study, uh, of the journal of healthcare innovation, highlighted the importance of interdisciplinary collaboration. It also emphasized continuous learning

Hanh: 00:09:31
opportunities and drive positive change within the healthcare organizations. So now what, um, strategies do you employ to create a synergistic environment that foster innovation?

Harvey: 00:09:46
You know, the number one thing is I always tell everyone that works with me, there are no stupid questions and there are no stupid solutions. We just need to work together. I don't want pride to hinder. So I have an open door policy. I want them to always to be able to come to me. Um, I love this little gadget. It's the Apple vision pro in continuing education. This is a big one. Uh, and this is a crazy story and not many non physicians know this story.

Harvey: 00:10:13
So in med school, the rule is see one, do one. And then teach one. And so, to give you a case in point, I was in residency, so learning how to be an ER doctor, and they said, Harvey, come over here, we're going to teach you how to intubate. It's when you put a tube down someone's throat and help them breathe, artificially. And so they showed me, and they're like, okay, the next one, it's your turn. And I'm like, oh my gosh, I mean, I've read about this, I've seen

Harvey: 00:10:39
it, and so literally I was doing it, and then as soon as I was done, they're like, okay Harvey, now your turn to teach the next guy, or girl. Imagine a world where I had this, and I was able to do this a hundred times. Over and over and over and practice and stimulate and over and over. And then when I did that first one with the real person, it really wasn't the first time. The muscle memory is there, the theory, my procedure. And so that's one way of how we can continue to improve our continuous

Harvey: 00:11:07
learning and create that supportive culture so that people feel like, Hey, this is the new way of learning.

Hanh: 00:11:14
That's true. And do you think that in the age of AI, continuous learning and lifelong learning and all of, you know, all in that realm is so important? And especially if you're going to keep up with the advancements of AI, right?

Harvey: 00:11:30
Yeah. You know, this is an interesting question. I call it the bell curve. We have people that I'll say here in I'm in Dallas, Texas. If I go to East Texas and I talk about a I immediately there's this wall and they're like, Whoa, whoa, whoa, that's the devil. No, no, no, I don't like that. No, that's that's not right. But I get the same talk in Silicon Valley. And they're the ones wanting to be like, yeah, high fives. Like, Oh, I like that. And it's a different thing.

Harvey: 00:11:57
So when it comes to healthcare and what we're talking about right now, we need to keep mind, like, what is the culture in that area? Maybe we don't show them the limousine. Maybe we show them a baby step, something smaller that they can say, okay, I'll accept that knowing that we can give them the limousine, but if we do, we may scare them away.

Hanh: 00:12:15
It's overwhelming. Yeah, no, I agree. You know, when I talk to folks. Sometimes they don't know where to start or what AI or how it can, how it can help them. My, my advice to many is like, start small, identify what your pain points are. I mean, you got to know your business. You got to know who your customers are and what goals you wanted to achieve. And then when you know that, identify what are the hurdles? What are those inefficiencies that you just love to do away and then prioritize,

Hanh: 00:12:44
start with a small gain confidence. I think once you experience that. And then you can dial things up. Uh, it is overwhelming, you know, for folks, yourself and me, like we immerse ourselves every day. So it becomes our vocabulary and the air and just everything in our, in our realm. So I think it's, it's, uh, it's so true that you, uh, you kind of test the temperature or the culture of the group that you're, you're talking to. So thank you. Now, recent study has shed some light on the complex, um, the complex of ethics.

Hanh: 00:13:24
Of using AI in health care and study in the Journal of Medical Ethics looked at the potential risks. of AI algorithms, if not, uh, developed and implemented responsibly, these algorithms could perpetuate biases and disparities in patient care. And the study also stressed the importance of addressing fairness, transparency, and accountability. So, considering the multifaceted ethical landscape of, uh, AI in healthcare, so how do you navigate these considerations? And what steps do you take to ensure that the technologies that you advise

Hanh: 00:14:04
align with the key ethical principles?

Harvey: 00:14:07
Yeah, no, this is such an important question, and I actually appreciate you taking time to ask that. You know, when I think about it, the number one thing that I think of is lack of education. If I'm a doctor, and let's pretend I'm brand new, I just graduated, I may believe everything I see. I may believe, uh, ChatGPT or some GPT that says, Oh, this is it because in my mind, I grew up with it. I've seen it. I'm thinking, Oh, this is the current standard.

Harvey: 00:14:35
Someone like me that's been doing this for 20 years. If you give me an output from GPT, I'll look at him like, Oh, no, that's wrong here. Here's why, because my clinical gestalt tells me otherwise. That's why we need to keep a human in the loop. So to answer your question. I think we need to educate. We need to educate everyone. What are the good, the bad, the unknown. And when it comes to ethical issues, some of the physicians may not

Harvey: 00:14:58
realize this is biased information and they're being told biased information and they're going to continue that biased information and reinforce it. But if we educate these physicians and nurses and health care professionals to say, Hey, keep an eye open for these things know that this is could be wrong. And here's why. The other thing that I think it's so important. I really want this to come out. We go to the grocery store. We buy soup. We buy something.

Harvey: 00:15:23
We look at the cans. We see the food label. Why are we not having a food label for large English model? Why don't I have something that says, Hey, Miss Brown, here's this A. I. Here's the food label. But you know what? It wasn't trained in this data. It wasn't trained in this. It's biased in the following things. And so when you use it, make sure that you keep this in mind. And so then now you look at

Harvey: 00:15:44
it, you're like, Oh, okay, I shouldn't use this tool for this. Maybe I should use this other tool. And I think that will help us create those right ethical frameworks. The other thing is we need transparency. That example that I gave you with GraphGPT, if I don't know how you got to that conclusion, kind of like in grade school when we do math in our head and our teacher gets on us like, no, no, no, show me the work. Same thing. Let's be transparent. Show me how I got to this.

Harvey: 00:16:12
What database did it go into? Where did you get that information? We're using PubMed. We're using recent studies. Did you confabulate this thing? Is this hallucinating? And so we need to be as transparent as possible. And we need to make sure that all the stakeholders involved, including our patients, are aware of the good, the bad and the unknown. When I consult hospitals, I'm telling them, Hey, we need it. Yeah.

Harvey: 00:16:33
A bill of rights, but like the bill of rights for AI at your hospital because patients are going to use Dr. Google now they're using Dr. GPT or Dr. ChatGPT.

Hanh: 00:16:42
Or Dr. Perplexity, right?

Harvey: 00:16:44
Yes. Dr. Perplexity. Oh.

Hanh: 00:16:45
Wikipedia.

Harvey: 00:16:47
Yes. And so we need as a society. To make sure that all the stakeholders understand what they're doing. I say this often, I'm, I'm worried for the people here in the United States, but I'm even more worried for people outside the U S that can self diagnose, walk over to the pharmacy and say, Hey, ChatGPT told me I need this drug. And they start taking it and there's wrong on so many levels. So I think we need to educate to make sure that these ethical guidelines are there for all of us.

Hanh: 00:17:16
Absolutely. It has some huge implications, but do you, do you see that? You know, AI in my mind, how I internalize this, it's a, it's like a GPS, right? But it still needs a driver and that's you, that's you and I were the driver. We, and it's all in our prompt, right? Um, we steer it, it gives us, it uncovers insights, uh, but it's, it requires you and I to prompt accordingly.

Harvey: 00:17:43
Yeah. My favorite phrase to what you just said is the way I see it is human, which in my case is doctor plus AI. It's going to beat out the best human out there and the best AI out there. And so let's be that GPS, let the human in the loop, but then I'm a human at two in the morning. Am I met best? Am I ready to go? Will I make mistakes? Possibly more, but can I afford mistakes? I cannot afford any mistakes, but if I augment myself and I use AI and I

Harvey: 00:18:12
use my clinical gestalt and together, I'm going to give such better care. Not that I'm stupid, not that I don't know, but at the end of the day, I'm human. But if I have these algorithms built into my electronic medical record, built into my workflow, now it's going to be better. Um, I'll show you this quick prompt. I got these uh, metaglasses and right now it's connected to ChatGPT, so if you ask me questions, it could hear it. Tell me what I need to say and they'll tell you, but in a clinical case, I could take a picture of an x ray, analyze

Harvey: 00:18:41
it, it'll tell me what it is, what it's seen, and so this is the type of stuff that will become common for every day. The other thing, um, this is a 3D model that I printed, it's the rabbit that's coming out, and they're saying, yeah, it's just like an app, but just work with me for a second. Imagine someone's having a heart attack. And I need to call the cardiologist, the hospital, the calf team, all these people. But if I could take a picture, obviously, make sure that's HIPAA compliant. And it automatically, I say, hey, code a heart attack.

Harvey: 00:19:12
And it automatically sends texts, pictures to the right people, to the right algorithm. And within seconds, I can still work on the patient, but that part of the work is done. And so that's what I'm saying. Like, let's use this technology and see how we can help people.

Hanh: 00:19:26
Absolutely. I love it. I love your thinking. You're innovating and encouraging the listeners to really take it to heart and proceed it with enthusiasm, but with caution and bring the good out of it and serve and solve more problems. So now, can you share a specific example of how AI has directly improved patient outcomes, directly?

Harvey: 00:19:54
Oh, yeah. No, this is awesome. There's so many that come to mind, but I'll focus on two. Since I'm an ER doctor, I'll share this one. One of my colleagues has this AI device and technology, and I'm jealous because our hospital doesn't have it, but here's what it is. Say someone's having a stroke. The AI is able to in real time, read the cat scan and send text to the radiologist, the doctor in real time saying, Hey, this person's having to bleed.

Harvey: 00:20:18
I'm reading the extra myself or the cat scan and then automatically whatever the radiologist is doing and saying, Oh, my gosh, I need to take care of this because it's showing alert. And then me in the front lines in the emergency room is telling me, Hey, This is, uh, a stroke. And so now I can start the ball rolling instead of waiting. You know, the radiologist has a list and then possibly missing it. That's one that immediately I'm going to be helping patients. Um, the other one that comes

Harvey: 00:20:44
to mind, uh, one of my friends invented this is pretty cool. It's basically a, a large language model device. That sits over the patient's bed, and it's just looking, and it's basically for the elderly. And so, as we know, if, God forbid, an elderly falls and breaks their hip, their mortality goes up. I mean, multiple things, especially, um, there's just a lot of stuff that hit their head. What this device does is in real time analyzing, and if the patient is making

Harvey: 00:21:10
a gesture, It can analyze but if it's ripping the IV or making the gesture of getting out of bed Then the ai is immediately texting the nurse texting the doctor saying hey your patient's about to get out and then simultaneously In their language using ai and saying if it's in spanish senora garcia, don't get up And so then it's talking to them and then and then i'm like wow, this is gonna save lives this is gonna be amazing because You know healthcare is expensive and I know hospital administrators as a formal one I used to pay people to sit by a patient to keep an eye on them now having

Harvey: 00:21:44
a device like this would be amazing.

Hanh: 00:21:50
Absolutely. But you know, um, here's another, um, consideration. Integrating AI into existing healthcare systems come with, comes with, um, significant challenges. There's the technical, there's the operational and cultural barriers that healthcare organizations face, uh, when adopting AI technologies. So, these challenges include, uh, for instance, data quality, standardization issues, interoperability concerns, and resistance to change among healthcare professionals.

Hanh: 00:22:28
Very common things, right? Now, there's also concerns about the potential impact of AI on the workforce dynamics and job roles within the healthcare organizations. And when you, uh, try to move this to, um, ensure a smooth transition and buy in from all the key stakeholders at all levels, it's so important for the successful of AI integration. So given the complex challenges, um, that you've discussed and I, uh, just added, so how do you approach the integration of AI into existing healthcare systems? And what strategies have you

Hanh: 00:23:08
found effective in overcoming, uh, these barriers and ensuring successful adoption process?

Harvey: 00:23:18
You know, you're going to laugh when I say this, I, I, I, my analogy is equivalent to arguing with your spouse. At the end of the day, how do you fix this? The solution or what's the solution? It's education. You educate each other on how you see it and you find that compromise. Same in health care. Someone comes in with a shiny object and say, Hey, we want to put this in your hospital. It's about educating the team. To your point, helping with the culture.

Harvey: 00:23:45
They're going to be resistant to change the interoperability issue going in and having someone like us health care that we know it. And when we integrate these things, we know that ahead of time that this has to be compliant. This has to be able to work with the electric electronic medical record. If that's what I'm trying to do. And so some of the other solutions. People are scared of this technology and they're worried that this is now going to be another cyber security issue. So one of my suggestions is

Harvey: 00:24:12
saying, you know what, I get it. Let's create a large language model and let's use something like LLAMA3 and let's put it in your hospital servers. That way you're in control of the data. It's not leaving. Your servers is staying inside your servers. And that's one idea that hospitals are very receptive saying, Okay, I makes me feel better knowing that this data is here because they know that if that data leaks, Oh, my gosh, you know, with HIPAA issues that could be literally millions of dollars that are

Harvey: 00:24:39
at risk if they don't do this correctly.

Hanh: 00:24:45
So what you're encouraging them to do is to have their own AI ecosystem, um, their own operating system, okay, where they can ingest their own data and not have it leak, but using whatever model applications out there. For me, it's Microsoft Azure AI, but there are others. I think that's key, right? To ensure, give them that confidence that is secured, private, and transparent. I mean, to me, that, that should be the first thing when you talk to folks who are hesitant about AI is to give them the confidence of their, their, their customer

Hanh: 00:25:23
service, their sales and marketing, their patient data, and so forth. So that's great. So now what excites you most about the potential of AI in revolutionizing health care delivery?

Harvey: 00:25:36
Gosh, my brain just exploded with all of that. Um, I get excited about a lot of things. Number one, seeing the ability of certain health care professionals sustained a profession because of a I makes me excited. Let me explain. If I know a physician or nurse is going to retire because they're tired of the system and let's say they don't like the documentation and let's pretend that we use AI to, but this conversation we're having is transcribing and it's going and now

Harvey: 00:26:02
they're like, wow, I don't have to type. Man, that makes me excited because now I know that with this aging population and aging workforce that we have, we're gonna get to a point where we're already there. It's gonna get worse.

Hanh: 00:26:16
Yeah, we're there.

Harvey: 00:26:17
Yeah. So if we can leverage a I to help decrease burnout. Help, um, job satisfaction with our healthcare professionals, then man, that's a huge win. So, to me, that's number one. Number two, obviously that's helping with efficiency, helping us with repetitive information, uh, things that we don't like to do. Give it to the ai, let AI do it. Mm-hmm, and, and AI can do that. The other part that makes me excited as a doctor is the predictive analytics.

Harvey: 00:26:43
I'll give you the example of New York Tron where I'm about to discharge. You pretend, and the AI is going through your medical record and says. Ms. Brown, there's another 20 patients that are, fit her profile, and if you send her home today, she's gonna come back by tomorrow in the next 10 days. And then to be able to use predictive analytics and say, you know what? My clinical gestalt missed XYZ. You know what? The AI is right on this one. I am NOT gonna let her go home yet.

Harvey: 00:27:12
I need to make sure I address these factors, and then I'll let her go. I was like, wow. I'm helping her. I'm giving her better care, and so that makes me really excited. And then the last is on the personalization part. Man, AI can take huge information that our brains can't handle. If I did a genetic makeup of what you are, and in my brain I need to figure out what your genetics and what diseases, I mean, that's a lot of information. But if I could put it into AI, And

Harvey: 00:27:37
help me create a better personalized medicine for you than man. That's amazing.

Hanh: 00:27:45
Exactly. When it comes to older adults, it's very personalized just because you have dementia doesn't mean your treatment is generic for all folks with dementia, right?

Harvey: 00:27:55
Correct.

Hanh: 00:27:56
And I love what you're saying is that even though you're using AI, but the key of it is you're using it to enhance or augment. Yourself, right? So it's to improve yourself. But nowhere does it replace your skills, your expertise and your experience. It enhances it. That based on what you explained. So that's awesome because I think that's one of the fears, right?

Harvey: 00:28:19
Yeah. Its. It is. And that goes back to my bell curve analogy. Someone that has had 2030 years may have a hard time realizing a I know some of this. And that's why I'm thinking not to take away from that person or her him or her. It's a combination of the two. But let's leverage, let's leverage those 20 30 years. The other parts of this is at some point that person is going to retire. But if I can create a digital twin of that

Harvey: 00:28:43
doctor and literally download their brain and put those clinical gestalt into an A. I. Algorithm. Now that carries on to the next generation of doctors that can take that wisdom that's not going to be found in any textbook ever. And that's what I will bring to the table.

Hanh: 00:29:01
Absolutely. And you know, for the users that have been, uh, prompting in their AI system, whether it's ChatGPT, um, CLAUD, or, you know, whatever applications, all the prompts that you are instilling, typing away, in, in some ways, you are creating a digital twin of yourself, right? Yeah, people don't realize that, but that's exactly what they're doing. And there's nothing wrong with that, because due to that digital twin that you may not be aware of, it enhances your creativity, your productivity, and even your critical thinking.

Hanh: 00:29:40
Because for every prompt and output that you get, You have to take it to the next level. Whether you want to improve the output, and it requires you to use your critical thinking. So now, how has your experience leading to leading health care systems influence your approach to a adoption and health care management?

Harvey: 00:30:07
That's a good question. You know, I feel that having the understanding of medicine, understanding what it is and being in the front lines Has opened my mind to be able to really think outside the box. Let me give you an example. At one point I had my own emergency room that I owned. And one of the texts is doc, why are we going to the patient room when we know that child's broken ankle or arm? Why don't we just take them straight to get an X ray and why don't we medicate them back in the X ray room?

Harvey: 00:30:35
And I'm thinking you're right. We don't have to wait for a label to take care of a patient. Let's do that. And so having that experience of medicine has opened my mind to say, let's think outside the box. I'll show another quick props, and it's this little plushy, and it has ChatGPT. And let me explain as an administrator, knowing medicine and knowing that this technology exists. If I have a child that I have a hard time communicating, why not use the plus sheet to walk them through a scary

Harvey: 00:31:04
moment to explain to them what the MRI is going to do or what's going to happen or speak to them in their language or in their voice of a five year old. That kind of information has helped because I know the problems as an administrator. I've seen it as a doctor. Now, putting those problems and looking at AI and how fast it's moving in some solutions. I'm like, What is wait a second, let's put this technology with these problems and let's create solutions to improve health care.

Harvey: 00:31:32
And that really has given me the edge of being able to have that leadership inside that strategic vision. Um, better allocation of my resources because I'm like, Wait a second. AI can actually do these things that you're not looking at. And so it's giving me an edge for sure. And I feel like all health care professionals will have that edge.

Hanh: 00:31:51
So true. And the plus that you just described, talk about personalization to your client, your patient, right? Now you're speaking to where they are, their language, their age level, and the condition they're in, whether it's, you know, Um, you know, they're there for an MRI or something. So true example of, um, relatable. Well, that's awesome. Yeah. Now, assessing the success and the effectiveness of AI in healthcare, uh, is so important to ensure that the

Hanh: 00:32:23
value, um, and the impact is worthwhile. And there are multiple metrics and frameworks for evaluating AI performance in clinical settings. Some of the metrics, uh, could be diagnostic accuracy, patient outcomes, operational efficiency, and cost effectiveness, right? And factors, uh, like user satisfaction and workflow integration, the AI's ability to enhance clinical decision. Um, are all, um, variables. So now, developing a comprehensive and standardized metrics for assessing AI success is very challenging.

Hanh: 00:33:06
Because different healthcare settings and AI applications may require tailored evaluation approaches. So given the importance of measuring AI effectiveness, So what specific metrics do you use to assess the success of AI implementations?

Harvey: 00:33:23
You know, the, the main thing I look for is I look at the team, you know, do we have a cross disciplinary team or do we have a healthcare professional? Do we have a AI specialist? Do we have an administrator? Are they all working together? And then the second that I look at is, are they communicating? Is the right hand talking to the left hand? Are they really, and those are some of my metrics. Because if they're not regularly meeting or communicating, they're going

Harvey: 00:33:47
to start going in different angles. And they're going to start solving it differently. And then the last thing is, I look at, I do metrics where I'm testing the data. Is that algorithm, let's pretend it's supposed to fix problem X. I'm checking it. So a month from now, three months. Is this still fixing X? Or has the reinforced learning created something called data drift where now it's not solving that problem like it used to? And so we need to make sure one of my metrics is to make sure that it is

Harvey: 00:34:15
solving the problem that you bought it for, that you created it for. And that seems so simple and elementary, but at times, if you're not aware of these issues, it's going to start doing that. So all those things are so important once we're developing these AIs in our healthcare system.

Hanh: 00:34:29
Mm hmm. You know, um, I think when you say education, collaboration, synergy and developing that trust, I think that's just key in any environment. But particularly, um, the implications of health care is huge. And when it comes to, um, like AI and all the complexities and the negativity that potentially could come with it, everything that you uncovered is really key. So I think it's important that, um, these silos that we're in. Might occur in the healthcare settings. We have to remove those

Hanh: 00:35:04
silos and collaborate across. Now, do you, or how can AI be leveraged to address health disparities and promote equitable access to care? What's your take on that?

Harvey: 00:35:23
Yes, that's a tough one. You know, again, I'm going to use my ER example because I'm an ER doctor. Being able to take real time data and analyze it Is amazing. For example, if I'm in the emergency room and I see someone with diarrhea, I'm like, Oh, just another diarrhea case. But then if I'm using a I and it's networked with a bunch of hospitals and they're seeing certain cases, they can say, Wait a second. This is bad. X. This is bad.

Harvey: 00:35:51
Me. This is from this one restaurant. Now we're able to communicate. And so to answer your question is, I really think it could help us analyze the data. So data analytics, it could help us with resource allocation. So if I'm looking at my data and I'm saying, huh, this part of Dallas is underserved and we're seeing these type of cases usually from this area, then I can use a I to turn around and say, you know what? Um, we need more resources in this.

Harvey: 00:36:20
It may be there. Water quality. It may be something as simple as they don't. They need a bus route to take them to the hospital. Maybe that area code doesn't have a way to get to the hospital. They can't afford. They don't have a car. They can't afford a taxi. Maybe they could afford it. So then those things is using a I to say, You know what? This is the solution.

Harvey: 00:36:39
Creating these customized interventions. Another one is I'm Hispanic. I speak Spanish. Um, being able to give better discharge instructions, better information. So customize my intervention. So if we're saying, Hey, the iPhone will take care of it, but then my population doesn't have an iPhone, doesn't have a smart device. So then now I need to see how can I fix that? What can I do? And I AI will help me leverage and it will help me figure out

Harvey: 00:37:08
what it is that I'm doing wrong. How can I intervene? And then. That could be my solution. So if I don't speak Spanish, use AI to help me translate in real time. If I give discharge papers that are in English, let's create them so that they're in Spanish, but let's customize them to their culture. Um, someone from Spain speaks different Spanish than someone from South America and people within South America. So why not? Why create one discharge

Harvey: 00:37:33
instruction for that entire culture? Why not create different ones specific for that race, ethics, age, so that we can really talk to them?

Hanh: 00:37:44
So true. Now I'll use an example with folks with dementia because, because that's what I know. Um, dementia in Parkinson's is the, the, the, the, the environment that I live in, because my siblings and my, my parents, um, for those that don't speak their, uh, their language, their native language. And to hear their native language while they're having dementia, um, it's, it, it deeply resonates because it brings back, um, the history, the stories, and what they remember. And I think that's really important

Hanh: 00:38:15
to have the ability, um, to relate to them in that way, right? Because it's hard to read and to get them to register what you're saying. So the ability to personalize, whether it's cognitive decline or the language barrier, that's so important. Right. Yeah. Now, what, um, strategies do you recommend for enhancing AI competency and readiness among health care staff? Like what education, what and where?

Harvey: 00:38:45
Obviously, I am really big on education. So I think we need training programs, but not just anyone. We need health care professionals that understand medicine and a I, you know, a lot of these training programs. When I look at their curriculum, I can tell it was written by a non clinician and that they're teaching the clinician. So it needs to have a interdisciplinary effect where we have doctors, nurses and we're using. So we definitely need the right training programs.

Harvey: 00:39:09
We need to change our education and our continuing education and make sure that A. I. Is involved so that they understand. And again, it goes to my analogy of the bell curve. We're gonna have people like me that are early adopters that want to learn everything they can. But then you're gonna have people that are late adopters and you can't teach those two people the same. And so we need to make sure that when we're teaching, we're teaching them to their level and getting

Harvey: 00:39:29
them all to a certain level. And so what I'm saying is we need practical workshops. We need something that I can go in and, and see the good, the bad, the ugly. But then someone like me might get bored in the first 10 minutes. So we need to kind of, all of us educate ourselves at certain levels. So what I'm saying out there is, We need to make sure that you're following people in health care like I follow you on LinkedIn. I'm learning from you I'm learning from other influencers And so that's

Harvey: 00:39:56
what we need to do is that continual mindset of learning forever to continue to say, you know what? I'm learning from dr. Castro, but i'm learning from miss brown I'm learning from all these experts and we need to create that culture and we need to instill it into the hospital system So that we'll continue to learn

Hanh: 00:40:11
Absolutely. So true. I love that you keep saying the word learning and open mind, lifelong learning, because we're in that place right now. That's period. We are in a place that we must adopt this everyday learning, adapting, opening your minds and trying new things. And the thing is, when you try new things, chances are it may not work for you the first time. Because even earlier on, when I was trying to prompt Well, you know, I wasn't so good.

Hanh: 00:40:39
I did not get the output that I was looking for. So it's not the tool. It's not the AI. It's your ability or inability to prompt. But you got to practice and practice and practice and after thousands and thousands. You know what? I've increased my efficiency quite a, quite a bit because I'm better at prompting. So I think part of the learning is very humble to know that you may not get the results that you want to get right away.

Hanh: 00:41:06
There's humility and there's keep pressing on and a childlike attitude and that's okay and that's what's exciting about where we are right now. Yeah. So that's great. So now how do you ensure. The quality, security and privacy of healthcare data used in the AI applications you described?

Harvey: 00:41:33
Yeah, number one, uh, obviously data encryption. Um, I'm becoming really good friends with our CIO and those individuals to make sure that our data is safe. Um, that's why I gave that suggestion earlier. It makes sense to me to have that large language model. On our hospital servers that it's encrypted that we know are HIPAA compliant that we know that the there's strong access control so that not any user can walk up to any computer and just log in where those access

Harvey: 00:42:00
controls are strong in the hospital. We got to make sure that those are going on. We're Continuous monitoring of who's logging in and what data is using. You know, this is an old example. But when I was in med school, I had a really famous person that came to the hospital and all these medical students were logging in to that patient's record because they wanted to see why they were there. And so being able to have that strong compliance monitoring to make sure that we're Regulating

Harvey: 00:42:24
who has access to that data. And then where's that data going? Is it going to chance you PT? Is it staying in our hospital servers? Is that data secure? And so that's a whole other hour talk that we can have on data security, but that's a very good question.

Hanh: 00:42:39
Yeah. Well, thank you. Well, here's a big one. Burnout. It's a growing concern in the healthcare industry. Now, studies show it can lead to decreased shop, uh, job satisfaction and reduce patient care quality. It can even lead to higher rates of medical errors. Um, so as healthcare organizations seek solutions, some researchers have begun to explore AI's potential in alleviating physician burnout.

Hanh: 00:43:08
So now, do you think, or what's your take on AI power tools that could help streamline clinical workflows and reduce AI burnout administrative burdens? Um, could it support clinical, uh, making, uh, decision making processes? And, um, I mean, we all know, at least from the outside, and you see it more, uh, day to day, there's AI based systems for documentation assistance. There's image analysis, patient triage, uh, could save physicians time and effort. This, uh, would allow them to focus more on direct patient care.

Hanh: 00:43:45
But the role of AI in addressing physician burnout is still an emerging area of research. There's concern about the potential unintended consequences, uh, it includes increased technology dependence. Transcribed And a need for more training. So a lot of complexity. So considering all the potential of AI in this context, do you believe that AI can help alleviate? Physician burnout. And if so, how do you envision a I being leveraged to support physicians?

Harvey: 00:44:18
Oh, gosh, I love this topic. So.

Hanh: 00:44:20
That's a giant.

Harvey: 00:44:21
Yeah. That's a big one. I love this one, though. So let's start with administrative automation. You know, if I'm speaking to you. And I'm catching your eye twitching, your subtle movements. Why not use AI to give me that information of this conversation? So it's transcribing it. But if my computer is over here and I'm typing and I'm looking at you every other sentence, I may be missing. A lot of information and and I

Harvey: 00:44:46
may I have a story of a physician that I hired and he couldn't type. He would type with two fingers and he would literally stay three hours after work and we would work 12 hour shifts and he would say 15 hours because he was trying to keep up with the paperwork. Why not use AI? And he was one that he's like Harvey, I'm ready to retire. But now with a I able to transcribe this conversation. Yes, you can hire another human to transcribe it. We didn't have that.

Harvey: 00:45:16
And so being able to have a I create these automations huge. The other one is mentioned earlier the decision support two in the morning. I'm a tired using a I to help catch things. Give me a better care for my patients. Amazing. We talked about the predictive analytics, you know, but the one that I it's a little controversy when I say is, I see a future where robots are going to come to play and I know people like oh my gosh, he's talking about robots I do I I see a future where say I had a robot next to me and

Harvey: 00:45:45
i'm taking care of miss brown And i'm saying hey, um, what is her hemoglobin? I want to see and based on your voice it's analyzing your voice and saying oh her sugar is this Or it's saying to me. Hey doc when you were examining her I noticed you didn't look at her ears and because of what she says, that's an important part of the exam. You need to go back and check her ears. And I'm like, Oh, thank you. Or what about my, my nurses that are really have bad backs because they're lifting all these patients. But if I had a robot to

Harvey: 00:46:16
say, Hey, help me with Ms. Brown, she needs to go to the restroom. Can you assist her? And now my nurse is like, you know what? I'm not going to retire. The robot is doing all the heavy lifting because physically I'm about to retire because I can't go anymore. So I really think this is going to be a huge solution that we're going to start seeing in healthcare.

Hanh: 00:46:36
Wow. Bless you. And I think you're right. And we just are excited to be a part of this, this day and age, right? It's so exciting to live and learn and grow along with the AI integration in our business life, but also in our home life, because it's like electricity in my mind, right? It's going to infiltrate every aspect. I don't know what part of life it's not going to affect.

Harvey: 00:46:58
Yeah.

Hanh: 00:46:59
Yeah. Yeah. So what advice would you give to healthcare professionals, um, that are wanting to navigate in this system? There's something going on with my Can you hear me?

Harvey: 00:47:20
Yeah, you sound great.

Hanh: 00:47:21
Okay.

Harvey: 00:47:22
Yeah.

Hanh: 00:47:22
I can't see you anymore.

Harvey: 00:47:24
Oh, I see myself, and I see it going so maybe she'll log right back in and so then I'll answer that question as soon as you get back. But to repeat it for everybody else that's able to hear me because I know we're live the question was what advice Um, would I give other health care professionals or students looking to contribute to AI health care? I believe is what she said. And number one, I would say, let's stay informed. Number two is let's develop our skills and then number three is let's collaborate

Harvey: 00:47:59
and then number four Is let's make sure that as we move forward, let's move forward ethically And so when I say stay informed follow, uh, miss brown here follow me follow other experts Um, if your hospital is resistant to change Then share this type of video with your hospital administrator. Be that advocate. If you're actually listening to this, that tells me that you're wanting to learn more. So pass this knowledge, whatever we gave you today to the next person next to you to improve.

Harvey: 00:48:25
And so let's collaborate together so that we can make a better workforce, a better tomorrow for healthcare.

Hanh: 00:48:31
Awesome. Well, thank you so much. I mean, I'm, I'm very blessed to be a part of this conversation, and I'm so thankful that we are in the age of AI, that we can share and we can learn from one another. And it doesn't matter where you are in this journey. We're students. We're students of, um, AI, of large language models, of all these applications that are coming out. So feel free to reach out, to follow Dr. Castro, to myself. Ask questions.

Hanh: 00:48:59
We're going to ask you guys, uh, gals questions as well. And to Dr. Castro, thank you for your passion and expertise and your vision for the future of healthcare is very inspiring. Your journey from practicing medicine to becoming a strategic advisor for GPT in health care is a testament to the transformative power of curiosity, innovation, and the relentless pursuit of better patient outcomes. Throughout our conversations, you've shared invaluable insights into the challenges, and And opportunities of

Hanh: 00:49:30
AI adoption from ensuring clinical alignment and data integrity to fostering collaborative environments and addressing ethical considerations. So your comprehensive approach to AI integration is a road map for success and your real world examples of AI's impact on patient care is very powerful. And your strategies for preparing health care staff. For the AI  revolution. Very practical and your commitment to leveraging technology to address health disparities and promote equity is admirable.

Hanh: 00:50:10
So these are all reminders. Of the immense potential that lies ahead into the folks that are listening, watching. I encourage you to embrace the insights shared by Dr Castro today and our conversation. So whether you're a health care professional, a student, or simply someone passionate about the future of health care, it's never been a moment more exciting time to get involved in the AI revolution. Imagine a future where AI and health care professionals work hand in hand.

Hanh: 00:50:47
A future where personalized, efficient, and equitable care is provided to patients worldwide. Where data driven insights guide clinical decision making. Where physician burnout is alleviated. By a I power support system and where health disparities are addressed through innovative technological solutions. Well, this future is within a reach, but it will take collective effort to make it a reality. It will require visionaries like Dr Castro, healthcare organizations, a I developers.

Hanh: 00:51:23
And individuals like you. So I ask you, what role will you play in shaping the future of healthcare? How will you contribute to the AI revolution that promises to transform lives and improve health outcomes across the globe? So the path ahead may be challenging. But with the guidance like leaders, like Dr. Castro and the passion and dedication professionals like you, we can create a healthcare system that harnesses the power of AI to provide exceptional care to every patient every time.

Hanh: 00:51:59
So thank you, Dr. Castro for sharing your wisdom and inspiration with us today. And thank you to our listeners for joining us to this exciting journey. So together, let us heal with data. One bite at a time, take care and thank you for your attention.