Leveraging AI Technology in Healthcare

In this podcast episode, Dr. Beverly Wright, Vice President of Data Science & AI at Wavicle Data Solutions, and Bhavna Mehta, Assistant Vice President of Data & Analytics at Cincinnati Children’s Hospital Medical Center, explore the transformative impact of AI technology in healthcare. They discuss various ways AI can support healthcare operations and decision-making, as well as the unique challenges faced when integrating AI into healthcare systems. Whether you’re a healthcare professional, a tech enthusiast, or simply curious about the role of AI in this critical field, tune in to learn more about how to leverage AI technology to enhance healthcare delivery. 

Speaker details: 

• Dr. Beverly Wright, Vice President – Data Science & AI at Wavicle Data Solutions  

• Bhavna Mehta, Assistant Vice President – Data & Analytics at Cincinnati Children’s Hospital Medical Center 

Watch the full podcast here or keep scrolling to read a transcript of the discussion between Beverly and Bhavna: 

Beverly: Hello, I’m Dr. Beverly Wright and welcome to TAG Data Talk. With us today, we have Bhavna Mehta, AVP of Information Services Data Analytics at Cincinnati Children’s Hospital Medical Center, and we’re talking about leveraging AI technology in healthcare. Welcome, Bhavna. 

Bhavna: Thank you, Beverly. It’s a pleasure to be here. I’m excited.  

Beverly: Well, let’s start off with a little bit of background on you. Why are you so cool?  

Bhavna: I guess because I’m not afraid to ask why.  

Beverly: Very nice. So, tell us more about your background. 

Bhavna: I am at Cincinnati Children’s Hospital. I’ve been there for almost 25 years now, so I’ve seen a lot of growth. I started as a consultant, but I always count those days as well. So, I’ve seen a lot of growth and a lot of change, both in terms of technology and the healthcare industry overall. I started in application development. I wrote applications to collect data and moved into application integration, like moving data between applications. Then, I moved into our quality department, looking at how data was being used to deliver care or improve care.  

And that was my first foray into really thinking about the potential and opportunity. Now, I’m back into IT but really looking more at data analytics overall and thinking about how we can leverage data to continue to transform care.  

Beverly: Wow, 25 years. I’m still stuck on that. That’s amazing. I bet the transformations have been incredible. Now 25 years ago, was it called Cincinnati Children’s Hospital? 

Bhavna: Yes. It was called Children’s Hospital and Medical Center of Cincinnati. They moved the Cincinnati to the front.  

Beverly: Cool. I’m super excited about this conversation because AI, in particular, is such a great opportunity and it’s a great time right now for improving healthcare. So, tell us about what’s going on in healthcare. 

Bhavna: You know, there is a lot of opportunity in healthcare, especially in the delivery of healthcare. I think AI has always been used in healthcare from a research perspective, so when you think about the huge journals of medicine, the publications, the discovery of new treatments and cures, we know that’s been happening, and a lot of data has been processed, including clinical trials, etc. 

The opportunity we have now is to shorten that cycle of taking it from bench to bedside that usually takes a long time because of technology not being ready or compute not being available. But with that, the complexity of data has also increased over time as we know with the technology evolution.  

So, now there is a lot more opportunity to think about in terms of predicting, say, trajectory of care or diagnostics or screening or operations even reducing clinician burden.  One of the things we’ve done with increasing technology and using electronic medical records, if you will, is that our physicians are probably spending a lot more time with computers. You would go to a doctor’s visit, and there are times the doctor is really entering notes into the medical record as opposed to spending time or looking at you directly and talking to you.  

There’s an opportunity to use AI as a tool to eliminate or reduce acclimation burdens so we can increase or give back human touch that care really needs.  

Beverly: And we can all relate to that because, when we’re sitting in front of our medical provider, he’s taking notes and not even looking at us or what seems to be not listening. So, it sounds like what you’re saying is that, on the research side,  AI and machine learning and data science, at least to some degree, have been there for a long time because of clinical trials, research, disease, and all sorts of things. But now we’re seeing it improve operations and, what did you call it, “bench to bedside?”  

Bhavna: Yes, bench to bedside. So, the bench is the research side. How do you take something that is discovered and put it into the clinical workflows? Historically, it has taken 17 years. It takes or has taken 15 to 17 years before where we are. But now, with technology advancements, we have the opportunity to speed that up. 

Beverly: So, the research that’s already been done for decades is now getting implemented much more quickly, and you’re also saying that operations, in general, are improving because of data science and AI. You mentioned scans, I want to come back to that one in a second, and the way that we’re capturing data on patients, and you mentioned patient experience. The only thing I didn’t hear you mention was equipment. I’m sure healthcare equipment is using quite a bit of data science and AI. 

Bhavna: Absolutely. Think about telemetry, patient alarms, ventilator devices, or heart monitors; all those are transmitting data, and we can use all that data to detect abnormalities, risks, or trajectories. So that’s a lot of data science going on, even in terms of alerting when someone needs to be alerted on an alarm, or there’s a raise, suddenly the vitals go up or something like that. So yes, there’s a lot of data science and potential to combine all that data, which now technology allows us.  

The other opportunity from a patient experience perspective is personalization. Healthcare has started becoming more consumer driven. If you think about it, a few years ago, even now, it’s not as open an industry. We don’t have that many consumer choices. But that is changing with small clinics opening up everywhere, so consumers do have more choices. There’s a little bit of consumerism.  

So, I think personalization from that perspective is going to be important. Usually, when other industries are catching up, people are expecting technology. A lot of it is like we, as consumers, when we are shopping at Amazon, have certain expectations. We probably want the same experience when we’re on a healthcare or hospital website. That demand is also driving how you would use it not only in the delivery of care but in operations and how we control the customer experience, the patient experience, and the employee experience, too. 

Beverly: Yeah, for sure. A couple of decades ago, you might walk in, and the doctor asks questions about your background. Now, if you did that to someone who’s under 30 years old and born in technology, they would say, “why are you asking me these questions? Don’t you have it right there?” There’s an expectation; that’s what you’re saying.  

And it elevates and maybe even pushes the envelope toward where we are going to implement, how we are going to implement, and how we’ve got to keep up with the demand. Also, to your point on the consumer, I had some medical experience and heard them use the term customer for the first time. Is that a thing now? 

Bhavna: It’s not. I mean, we’re trying to use those same learnings, if you will. Again, healthcare can learn from a lot of different industries. If you say, “customer,” it’s really patient-centric. The patient is in the center, and then everything revolves around them. That’s typically the terminology, but it depends on if you’re trying to learn and borrow ideas from other industries that could creep in. 

Beverly: Yeah, I was like, “why did that sound so strange?” 

Bhavna: I think we deliberately would try to say “patient” in the center. We want to be patient-centric.  

Beverly: Okay, cool. I would imagine that there are specific challenges. I told you I’d come back to something you mentioned about scans. Because on the surface, you listen to all these different things that are happening, and you think, “okay, yeah, I’ll take two. This is great; of course, I want to improve patient experience; of course, I want better operations and more.” How do we get more of this? It’s not actually happening everywhere in mass, it’s not happening at scale, and some of it’s not happening at all. It’s more conceptualizing. 

So, is there a little bit of fear, too? Because you talked about scans, and there’s a lot of talk about how AI can read scans better than people, and you talked about patient experience a good deal. What are some of the big barriers specific to healthcare that are keeping us from gaining all these benefits?  

Bhavna: Yeah, one of the things is data. There are different modalities of data that you have to ingest, and you can. Healthcare data is wide and shallow, too. The data quality is not the best, as you would say, after all, no two people are identical, so that’s a challenge.  

Data engineering, mapping data, and getting data ready for AI are huge things. We’re still working on that.  

Beverly: Every human body is complex, and gathering that data, and if you ask their opinions, a lot of patients lie. I read about 20% or so of patients just lie outright to their medical professional. So, there’s all these challenges with data itself. 

Bhavna: So, data itself, and again, obviously, data is biased toward people who come to the hospital. That’s also another thing. You’re already inherently biased toward sick people; that’s what it is, so data challenges.  

Then, of course, you’re making decisions for people. There’s fear. You want to make sure because we’ve all heard about things where machines can go wrong. I shared an experience this weekend. We tried the self-driving Tesla, and it was cool, but it was creepy at the same time. At that point, you think about the humans still responsible. It can do crazy things, even though it was doing a great job. It didn’t know the shortcuts, so we put an address in the GPS. Yes, it took us exactly, but there was a side road that we would typically take to go to that place. It doesn’t know that.  

So, I think some of those things, as you start to apply those to when you would put it into production in a clinical workflow, there are a lot of things we have to consider. One of the things that healthcare, in general, is looking at or thinking about is “how do you use AI as a tool to augment?” Not to take the hands of humans in the loop, but to use it for decision support. How does it make the human better and not replace the human? Essentially, it is the guiding principle.  

Bias is, of course, another thing. I have to make sure that we’re offering equitable outcomes, and in the data you’re training with or the algorithms you’re training on, there’s an inherent bias in society itself. We just have to figure that out and be more careful about it. I guess the severity of the outcome, or the impact of the outcome, is much larger.   

Beverly: Yeah, that’s a good point. So, it could mean community health, not just one person. This colleague of mine is great. He works in the western part of Georgia, and his number one data scientist had been a nurse. She was an RN and went back, and she said it was because she felt like she could help more people than one at a time.  

But that also means you can hurt more people. So, if things are done wrong or something doesn’t go like you wanted it to, it makes it larger; it makes everything bigger and more extreme.  

Bhavna: Yeah, there’s that. But on the flip side, there’s also a possibility to identify and mitigate bias that humans can’t recognize. So, it’s using AI for the right thing but not taking your hand off the wheel. Making sure that the outcome is aligned with how a physician would predict it.  

One of the metrics that we always talk about is integrated reliability. If you have a model giving you an outcome, and there are four people that’ll give you the same result, then yes, that seems like it’s in alignment with the education and the training of a clinician professional. There are different ways of validating something, yet still being skeptical.  

Beverly: Yeah, for sure. And that’s what I was going to ask you is: if you could envision how it should be, because we talk a lot about the human in a loop, especially in healthcare, and you don’t want the AI just making the decision, what would that look like in the right way with medical and healthcare decision-making that’s sort of augmented with AI? What does that look like? 

Bhavna:  Yes, some of the things would be risk protection. One of the metrics that’s often used in healthcare is: what’s the risk of a patient coming back to the emergency room after they have surgery? You usually measure 30 days on returns, and there are key parameters that would indicate that based on history or adherence to their past medications or protocol, etc. If we could have that indicator upfront, then you can actually do early interventions.  

So, helping and augmenting that would be an example. If social risk could be helpful, so if we can predict that in advance or know that in advance to say, “hey, this patient lives in a certain community or has been exposed to certain social terms of health, can we circumvent those sooner?” In some of those areas, it can really help with. Or even, images you talked about, so if it can give a preliminary diagnosis, then somebody has to write up, and that’s how it’s being used, almost like a preliminary diagnosis. Whatever you do from an imaging perspective, it looks like that.  

The other thing you could do is enhance; it’s a great way to enhance images. What helps is you don’t have to expose that much to those rays. You don’t have to go through X-rays again and again in that imaging. Because, even if it’s a poor-quality X-ray, maybe there’s an opportunity to enhance that without exposing the patient to more. There are some of those opportunities for sure.  

Beverly: I like that vision. That’s a good vision. Is there a fear of AI? Because we’re talking about creating models with AI and then you collectively use these models to make some final decision and the person is still using the decision. But is there a fear at some point that it’s going to take the driver’s seat or that it’s going to eliminate jobs? What do you think about that?  

Bhavna: In healthcare, I don’t think that’s there per se, because it’s still very early. Maybe on the operations side, if you think of the other areas, there’s some of that. In my mind, I don’t think that’s the right way to think about it, and it’s really used to think like supporting your buddy or an assistant. I do think we’ll evolve.  

The biggest strength of human beings is that we evolve. It is this change that takes time, and I do think that the way we want to think about it is that it is really a tool and can help us enhance what we do and not replace what we do.  

Beverly: Right. How many carriage riders or drivers do you know? I don’t know any carriage drivers. There was a big fear about carriage drivers getting out of the job, and now here we are.  

Bhavna: We’ll invent new jobs. That’s evolution. It’s going to take some time. But every time we have a disruption, that’s going to accompany fear, too.  

Beverly: What do you see as some of the very top types of opportunities? I’m sure you’ve thought of this in the community: is it anti-diabetes-type stuff, or is it like public health in some other way, or activities, or a different mindset of how we care for our own selves? What would you see if you had a silver bullet and could go, ”here’s where AI can really help us; let’s do this if we could just do this one thing.” Where do you see it? 

Bhavna: “Clinician burdens” is what I am hearing. Can we give back time to our patients; that human touch? Because that’s what my colleagues and whoever I’ve spoken to say; they come into medicine to actually help people. And sometimes, we create that burden that they’re not spending as much time as they would want to. That would be one thing.  

Of course, it’s hard to rate because there’s so much opportunity because it could really expedite research. We talk about operations, but it really expedites research. So I was listening to, I don’t know if it was NPR or Wall Street Journal. One of the opportunities on the clinical trial side that they were talking about was where we have this control group and another group that actually gets the medication in a clinical trial. The control group gets the placebo, if you will. But if AI could be used to predict the trajectory of a disease, then everybody could get the medicine sooner, a real thing, and you don’t need a control group.  

So, again, there’s so much opportunity to think about expediting and accelerating research, as well, that could lead to new cures. So, I’m not sure about the silver bullet, but I am passionate about informatics in that way. I love to see that time given back to the patient in some way and improve the patient and doctor interaction. 

Beverly: Right. Better, more improved, accurate, deeper research that gets bench to bedside faster is probably one of your top things. The second is reducing the clinician burden so we can give back time by increasing that human touch back to the patient.  

I love that. Very nice. Any final piece of advice about leveraging AI technology in healthcare? 

Bhavna: Taking it, leaning into it, not being scared of it, trying and starting with pilots, and studying—that’s the biggest barrier. One of the things is that either people are staying away or not going very fast because it’s scary. Something to think about is having diverse perspectives, so it takes a full team; it’s not just IT, but business, you’ve got legal, you’ve got risk, so we need a lot of different perspectives when we’re trying to implement something. Let’s make sure that our team has diverse perspectives before we implement something else. 

Beverly: Learn more, reduce fear, and consider multidisciplinary approaches. That’s very nice. I love it.  

Thank you so much, Bhavna Mehta, Assistant Vice President of Information Services, Data Analytics from Cincinnati Children’s Hospital Medical Center, for talking to us about leveraging AI technology and healthcare.  

Bhavna: Thank you, Beverly, it was fun. 

The impact of AI technology in healthcare is both promising and multifaceted. This insightful conversation between Dr. Beverly Wright and Bhavna Mehta highlights the various ways AI can enhance healthcare operations and support decision-making within a patient-centric healthcare ecosystem. Staying informed about these advancements and understanding their applications can significantly improve healthcare delivery.

Explore the full catalog of TAG Data Talk conversations here: TAG Data Talk with Dr. Beverly Wright – TAG Online.