This Week Health

Don't forget to subscribe!

February 9: Today on TownHall, Brett Oliver, Family Physician and Chief Medical Information Officer at Baptist Health speaks with Hamed Abbaszadegan, MD, Physician Executive at Stanson Health about AI, machine learning, and clinical decision support. Where in healthcare does he see the greatest opportunity for AI today and in the near future? How is clinical decision support evolving to be more useful to end-user clinicians? What would he recommend health systems do to routinely monitor these technologies? What is Hamad most excited about in the AI/machine learning space?

Healthcare needs innovative ways to address staffing shortages from clinical to IT employees. Are you curious about how technology can help support your Healthcare staff? Join us on our March 9 webinar, β€œLeaders Series: The Changing Nature of Work,” to explore how Health IT can be used to supplement Healthcare professionals.

Subscribe: This Week Health

Twitter: This Week Health

LinkedIn: Week Health

Donate: Alex’s Lemonade Stand: Foundation for Childhood Cancer

Transcript

This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

Today on This Week Health.

to come back to your MAP application, five years ago, you would almost always get to your destination faster than the prediction but nowadays, I feel that I rarely beat the time. better information is becoming processed better. And we have to build that same confidence amongst clinicians with our clinical decision support

Welcome to TownHall. A show hosted by leaders on the front lines with interviews of people making things happen in healthcare with technology. My name is Bill Russell, the creator of This Week Health, a set of channels dedicated to keeping health IT staff and engaged. For five years we've been making podcasts that amplify great thinking to propel healthcare forward. We want to thank our show partners, MEDITECH and Transcarent, for investing in our mission to develop the next generation of health leaders now onto our show.

Hello and welcome. My name is Brett Oliver and I'm a family physician and C M I O for Baptist Health in Kentucky and Indiana. And I'm happy today to be having a conversation about AI and machine learning with Hammed Abaza again hammed is a physician executive with Stanson Health, and I wanna welcome you to the show today.

Thank you so much for having me. It's an honor and appreciate conversations around machine learning and artificial intelligence.

Yeah. I'm looking forward to this. This has been on my calendar for a bit and I, I've, I've got several questions for my own personal information that I really wanna pick your brain on.

So, well, let's just jump right in. I, if you could just give us a little bit background about yourself and with ai, machine learning or in general, and how you got into this. Overall,

absolutely. So, always in forever we're clinicians first. First as physicians. So my background is internal medicine, hospitalist based medicine.

Is my clinical domain and really my inroads into this field. Really initially started with doing the chief residency in quality and patient safety. And from there I was a chief of a informatics department at a local hospital here in the Phoenix Metropolitan. And I pretty much was doing that for the last nine years.

Over that time use of data to optimize healthcare delivery really has matured and expanded. And with that, the techniques and software have pretty much followed. And if you really look around, you, there's so much machine learning and AI that has been around I mean from, fraud detection with every swipe you make of your credit card to going through algorithms to determine is this broad or.

To how Amazon leverages market basket analysis to give you suggested purchases based off of your search history. That is all data pattern recognition of what other people have bought, who've bought the goods that you were searching for. So as I always say, healthcare is so far behind everything else. I mean, I was literally handwriting my notes and document.

In:

Prescriptive type of analytical work can be leveraged and utilized for healthcare delivery. So that's in a nutshell, my background.

That's fantastic. Yeah. Exciting stuff. And you're right, we were handwriting way too long. Yes. So. Fast forwarding to today, where in healthcare do you currently see the greatest opportunity for ai?

That's, that's actually available that we can use, that maybe some are already using, but not everybody realizes that an opportunity's out there. And then as sort of a corollary to that, what about maybe three to five years from now, what's in the pipeline that you see that's gonna really be a, a great opportunity for clinic?

It's such a great question, and I think really the answer to start off relies on, in our own lives how we utilize AI and machine learning and analysis of data. So one of my favorite examples that I frequently talk about is the maps application on your, you know, iPhone and you're gonna input. in an address and it's gonna basically computate a whole lot of data points to give you a prediction on how long it will take you to get to a destination.

But it's also going to give you decision support for you to decide which route should I take? Should I take the highway, should I take the surface streets? Some degree of that. We have an idea of on every journey we. take But we use that data and information to make decisions. So where we're at today is at that point, we can make better predictions on disease processes, on predicting outcomes, on looking at predicting who might get cancer, who might develop lung cancer, for example.

And what we can do in the medical field is screen for that ahead of time. And so that therein lies a lot of where we're at today. Lots and lots of prediction, predicting who's not gonna show for an appointment. Predicting whose diabetes will be controlled, won't be controlled predicting cancer states.

And so really a role of what we play by leveraging this data and information is ultimately comes back to making decisions and utilizing technology to make decisions and support those decisions we make. I'm pretty confident it's almost across the board to some. extent This is being done across the country in different health systems, especially if you look at things that have been tried and trued as a big hot topic, like readmissions into the hospital, looking at patients meeting criteria.

So a lot of these different components can start to be automated and, you know, scanning for specific data points, predicting prior authorization, looking at front office back. office Making sure the right code is on the right patient, not upcoding, not downcoding capturing what the patient truly has that's already within their electronic health record and basically leveraging that information to make decisions.

So where are we going in three to five years? I think if you look at a lot of prediction, the biggest explosion in data is us, ourselves, the. and all of our wearables and devices. Okay? So tracking how many steps we're taking, tracking our heart rates, tracking our sleep patterns, looking at how well our C P A P machines are fitting and if we're getting good oxygenation at night, when we're sleeping, tracking data that our mattresses is recording in terms of what softness we want in our.

So all of this data is getting stacked along with looking at genomic data along with your personal health information, your labs, your meds, you're on, and computating more and more what I consider stacking of data sources and getting even better with our predictability of disease states, cancers, and health needs at the moment that we need them. πŸ“

β€Š πŸ“ It is:

You can get more information about them at alex's lemonade.org. How can you help? For the month of February, we will be holding a download drive. We're doing a bunch of different drives this year. , and our hope is to raise $50,000 for Alex's Lemonade stand and for February the download drive for every download over 20,000.

And just so you know, our average is roughly about 20,000 every month for this week. Health over the various channels. So what we're gonna do is for every download over 20,000 in the month, February, we're gonna donate $1 to Alex's lemonade stand. So if we get to 25,000, we'll donate 5,000. If we get to 30,000, we'll donate 10,000 to Alex's lemonade stand.

A download is counted as. Anytime someone listens to the episode of this Week, health on either of our channels on the conference channel or on the Newsroom channel let your staff know your peers, whoever you think might benefit from listening to our interviews and this content, and support the work of Alex's Lemonade Stand.

. Already raised in:

Go ahead and give a donation. Leave a little note. We'd love to thank you for participating in that and look, it's really easy. Shoot a note to somebody who you think would benefit from listening to this content. And for every download above 20,000 this month, we are gonna give $1 on your behalf.

So we want to thank you for all your sport and help as we try to give back this year. πŸ“ β€Š πŸ“

Yeah, I think you spoke volumes there at the end. There's all this data and there are times when it feels like as an end user clinician, that data is just piling up on me with no actionable insights. I can remember when we. Several years ago integrated Fitbit data into our E H R and, not to, not to knock the analyst that made it happen and all that, but it, and it was sort of celebrated and it's a good thing cause it was sort of the first step down another road, but it's not clinically useful, like to have just dozens and dozens of heart rates all you're doing from a clinician's perspective, this feels like setting me up for liability.

What if I missed that, one value that was there. So, Evolving that, clinical decision support, let's be specific about what we're talking about here, evolving that clinical decision support, how has that evolved? And, and I guess we touched on that a little bit, evolving to present, more complete picture, complete data.

But how has that evolved in advancing that for clinicians and, what do you expect to see there in the future specifically to CDS

Yeah, it's such a good, good point. And in fact, I chuckle inside because your Fitbit example is one that I would notoriously give when I would talk on this topic that yes, we can monitor your heart rate, but is it worthwhile to pursue that clinically?

I mean, of course there are circumstances of detecting certain arrhythmias. We're not getting into that. Randomly knowing random heart rates at certain times of the day, of course doesn't have clinical utility. So you know, where is this kind of headed? And I think that software machines, pattern recognition and ultimately AI can basically differentiate the noise and what's reality and prompt.

within Workflow of when clinicians are making decision in appropriate manners. That I think is where the key element is coming, because the computational power is better, the software is better. The way in which we use these tech techniques to differentiate things within records is ultimately where we're headed.

And so if I give you relevant information when you wanna make a decision, you're more likely to listen. But if I send you irrelevant information, it's gonna be tough for you to sort through. Probably we won't sort through it. And ultimately we lose confidence in those data sources and technologies To be very frank, so we're in this stage, I feel of building confidence and to come back to your MAP application, I would say five years ago, you would almost always get to your destination faster than the prediction you were given when you began your journey.

Okay? But nowadays, and this isn't my own life and own. I feel that I rarely beat the time. That's a very good point. What that, yeah. So what that tells me is the data points, the data analysis, and. noise Is getting less and less and better information is becoming processed better. And we have to build that same confidence amongst clinicians with our clinical decision support and how artificial intelligence is leveraged just as we've built the confidence we have in our directions we get and time of predictability when we'll reach our destinations.

Do you think, you know, one of the, banes of a clinician's existence is, I'm in the middle of my workflow and I get a BPA firing or something that stops my thought processes. It made me think of it as you were answering this last question. Do you think we're gonna get to a point relatively soon where the, AI is in the background and until I act, until I say, okay, we're gonna use this medication, and the AI then says, Hey, with this patient for these reasons, maybe you should consider an alternative rather than before I get a chance to do the right thing.

What are your thoughts there? Does that question make sense?

It makes sense to me. Yes. And I feel that. Farther along than individuals might realize. I mean, getting the right information in the workflow of the clinician is basically a core principle of, what I'm involved with in the, development side and the support.

I provide that with Stanson Health as well as my other colleagues. But to add another component to what you mentioned, I mean, pharmacogenomic data will predict our metabolism to certain drugs. So if that's built and integrated within an electronic health record as some sort of clinical reminder, order check.

When you go to order drug X. But we know that that person will have a bad reaction based off their pharmacogenomic lab data. Then Option Y can and will be prompted. And this type of work is already in play in health systems. It's only gonna proliferate. It's only gonna expand. It's only gonna get better.

And it's gonna prevent you and I and , our children and our family members from getting the wrong drug based off metabolism and genomic components of how we process things. And that's what's really exciting. It's beyond just leveraging, the technology for predictions. It's also making sure that.

The right data is used to make sure the right meds are given. I think that that kind of ties into what you were asking, but how that that's gonna play out. I've seen it in play and again, , it's only gonna expand.

Yeah, I love that. I think when you can support the clinician, support the nurse, but wait until they need it, rather than just this more universal, blunt instrument that's exciting.

That's just, that's why a lot of people say augmented. Not artificial, because ultimately it comes back to augmenting, to supporting your decision making, never ever doing the decision making thing.

Yeah. Very, very well said. Well, along those lines, you know, one of the things that we've been challenged with lately, and I, and there are probably some that are more advanced in this spot, but I'm finding a, real problem, understanding how a health system, a true artificial intelligence or augmented intelligence, A true algorithm like that that has the real potential for drift and things along those lines.

And not to get too technical, but how should a healthcare system evaluate these things on the front end? And then what would you recommend they sort of set up to monitor them as they go? Cause I feel like sometimes these things get put into place and then, We never think about it again. And you know, at least here, we're trying to evaluate things on a periodic basis so we make sure it's still doing what it's supposed to do.

Yeah. Such a good question. So important. You know, So many different companies out there making claims on how tech can be leveraged and use our software. So how do you really differentiate that? . And I think the answer goes back to how do we really differentiate things. Before technology, we always relied on our societies.

American College of Physicians, for me, on the internal medicine side American College of Family Physicians, American College of Radiology. So the different societies, the different universities, academic based centers are usually the ones that come out with guidelines. And sometimes it can be at a federal level, f D, a, c, D, C.

And there's a fair amount of high level confidence in these institutions that they're unbiased and do their research and make sure that these guidelines are created based off of evidence-based research. And so now you want to, tech that, up or. Electronically convert those guidelines into a software is in essence what certain companies like Stanson Health does.

And so the answer in that is understanding the logic, leveraging your informatics personnel and physicians to. Help understand and evaluate the logic of what goes behind looking for what types of data and what recommendations are being prompted, and is that following the latest evidence-based guidelines that are, put out there by the universities and C d, C, and are these types of things approved by governance bodies?

I think for me, I still look to those institutions and I mean, we all went through the pandemic recently and I mean, I can tell you from my perspective, I was always watching the cdc. So same things. Look to the leaders and leadership behind evaluating evidence-based criteria, algorithms, diagnoses, what you're supposed to do and not do.

And then companies that leverage that and work directly with some. Guideline based developers I think is key. πŸ“

β€Š πŸ“ If you haven't heard yet, we're doing webinars a little differently this year. We got your feedback. You wanted community generated topics, not vendor generated topics. You wanted great contributing panelists. Definitely not product focused, more focused on the challenges and the problems that we are facing in healthcare.

We are only making these available live. So we are making them more dynamic in nature and we're doing them on a fairly consistent time, as much as we possibly can. The first Thursday of every month. The next webinar is going to be on March 9th. Which technically is not the first Thursday of every month, and I apologize for that, but I'm actually on vacation that day.

So March 9th is gonna be the webinar and we're gonna continue our leadership series. We're gonna be talking about the changing nature of work and a lot of things have changed. The pandemic drove us to work out of our homes. What does that mean? What does it look like? How are we making decisions?

Are we making data driven decisions on that? How are we maintaining culture? How are we hiring? Are we hiring differently? , and not only that, not only focusing on it and the roles there and the challenges there, but also on the challenges that our health systems are facing. The changing nature of work as we move into working at hospital, at home, , and some of these other care venues.

what does that look like? Addressing the staffing challenges in the clinical side as well as the administrative side. So, we are looking forward to having that conversation. Love to have you join us March 9th. Keep an eye out. We're gonna announce who the panelists are gonna be. I currently have my feelers out for some people, but you can count on the fact that we're gonna have great panelists.

We're gonna have a great discussion. You can sign up on our website this week, health.com. Top right hand. The cool thing about that is you could put your question right in there, and I give those questions to the panelists ahead of time and we make sure we integrate that into the discussion. So sign up today, hope to see you there. πŸ“ β€Š πŸ“

Gotcha, gotcha. Makes sense. You know, One of the things that I, ponder too is if true AI is used to help support care of a patient, at what level or what understanding does a clinician need to know here this, you're being presented with a best practice advisor, let's say a bpa.

Was there AI used in that or was this just , simply, you know, some very simple math, not anything that changes with time, and then does a patient, is there transparency? A patient needs to know? We're making this decision based on some ai and I, I don't, I don't have an answer to that, but I, I'm curious to your thoughts.

I think that what you're alluding to, if I'm correctly, is the governance component of this. So your own internal governance to approve and not approve certain technologies or certain types of advisory alerts. What is the process for that? And I can tell you we actually have a governance playbook that we suggest to different health systems we work with.

But you know, moreover, we participate and help evaluate those as well. and the best governance groups evaluate the evidence from literature and evaluate logic and what is goes into making these, you know, inferences or suggestions to our clinicians. Sure. And I think the highest performing health systems have a good process to look at these internally, leverage their partners, but make sure that that logic.

that is in play is, clinically sound before we start making suggestions to our clinical staff making decisions for patients.

Yeah. So let's say your governance body approves a particular clinical decision support algorithm. Has it been your experience that you've seen then when that fires, I get the approval process and making sure that it's f d a approved , and meets certain criteria, but.

Is there an obligation to let the clinician know when that CDs pops up to say, this is generated by ai, or is this gonna get folded in with other clinical decision support? Recognizing that it's governed by your organization in certain ways. Does that make sense?

Yes. It makes better sense. And I think what you're getting at is do you wanna tell the person that you're alerting what the clinical details are and what the sources and reference are? My answers absolutely. Yeah. To suggest it was AI or machine learning. I mean, in my opinion, I don't know that that's needed.

I mean, most people understand Amazon and it's all machine learning and ai. I don't know that it would change your opinion. , I'll be very honest. When I do a search on Amazon, I very rarely order the suggested parts. Sure. I make my own decisions based on my needs. And so flipping the script to a clinical.

You can put that in there. But I think what clinicians typically really want is they want to know how did you come to that? And that might be as simple as suggesting, Hey, here was the last lab to, back up this inference we're making that this patient has a high prolactin level and you should code as that because they're ProAct level was this amount.

Yeah, two months. .

That makes sense. Yeah. It'll be interesting to see how it evolves, because on a lot of our alerts, we'll put, like, we have some pharmacogenomics integrated and the clinician if they want to, I haven't looked at the, recent numbers, but if you want to, you can link out to what this is based on.

Yes. I wonder if with time, , they'll develop some trust in, the systems and things. , We'll, you know, we'll see verdicts out on that one for now, but

I think over time we. I trust my MAP application more. Every time I feel like, oh, that's wrong, I'll. , I'm starting to lose now. I'm starting to not be as good as, the suggested decisions.

I'm, being prompted.

I love the map analogy. You're exactly right, though. Especially on longer trips, eight, nine hour drives that I'm not familiar with the area, and it says I'm supposed to get off here. Hmm. I don't know. I don't know. But to your point, over time I realized now it was right.

This is the faster way to go, even if it doesn't, seem right to me.

and one fun point to bring about that is , I don't wanna cost compare companies, but there are certain MAP applications that we have more confidence in, that work better. And in my experience, it's the ones that have more data input and have been around.

So I think that computational power is there for everybody, but the amount of data that is being computated is gonna vary. And so confidence comes in those predictions being more and more accurate to true life and true form. . And so I think as time progresses, that gap will narrow.

That's a great point.

Yeah, exactly. Well, let's wrap this up. One more question for you and then I'll let you get, I can't believe Sure. Time's already up, but just basically I wanna open it up and say what's got you most excited in this space? I mean, we've talked about a lot of things and maybe it's something you've already mentioned, but I always like to, ask, you know, what's got your most.

I'm the most excited about connecting the dots. So we talked about a lot of different points here, pattern recognition, predictability, knowing who won't show for an appointment, and leveraging different types of data, looking at our genomic data, our metabolism, our personal data that we've accumulated from wear.

So this data stacking and computating that to get us the most accurate information to predict things for us that we can take action on. Okay. That's what's most exciting to me. So we all have known of some family member friends that were diagnosed with bad diagnoses at early age. And I believe that those are gonna become less and less over time because as we stack and compile different sources of data that are true and specific to an individual, then the output and prediction and knowing what will happen is only gonna get better.

And if you know what's gonna happen ahead of time, You're gonna take an alternative route. If you know there's an accident ahead of you, you're likely to take the detour. If you know you're gonna get a disease, state or lung cancer, you're gonna screen so that when it imminently pops up, We go to the right doctor at the right time to excise that tumor and prevent us from frankly dying.

So I think it's this connection of this technology and data , that's most exciting. And one other comment, one of my colleagues always told me is optimization is a forever event. So these concepts we're talking about and getting better and making these predictions better , and leveraging data and information.

This is not gonna be done in five years and then we're done. It's only gonna improve and it's only gonna be applied to new disease states. New frontiers and the problems of tomorrow that we don't know will be able to accelerate answers and analysis to make better decisions for those problems. Just as we saw with the rapid development of vaccinations and other types of things that we've seen in modern day.

Yeah, it is, it's really exciting. , I wonder when the day comes, and I don't think it's too long to wear if you're not using. AI machine learning in certain aspects of care. It's malpractice because of Yeah. Advantage that it gives you in care and it's pretty exciting to see. So, I really appreciate your time and thank you for being here with us and I certainly learned a lot and would like to continue the conversation.

Absolutely. My pleasure. Thanks for having us and thanks for having me. And look forward to also talking in the future. More. Absolutely. Take care.

gosh, I really love this show. I love hearing what workers and leaders on the front lines are doing, and we wanna thank our hosts who continue to support the community by developing this great content. If you wanna support This Week Health, the best way to do that is to let someone else know about our channels. Let them know you're listening to it and you are getting value. We have two channels This Week Health Conference and This Week Health Newsroom. You can check them out today. You can find them wherever you listen to podcasts. You can find 'em on our website this weekhealth.com, and you can subscribe there as well. We also wanna thank our show partners, MEDITECH and Transcarent, for investing in our mission to develop the next generation of health leaders. Thanks for listening. That's all for now.

Thank You to Our Show Partners

Our Shows

Related Content

1 2 3 295
Healthcare Transformation Powered by Community

Β© Copyright 2024 Health Lyrics All rights reserved