SCTS_Judy Murphy .mp3: this mp3 audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.
Intro/Outro:
Welcome to the Smart Care Team Spotlight presented by care.ai, the smart care facility, platform company, and leader in AI and ambient intelligence for healthcare. Join Molly McCarthy, former CNO of Microsoft, as she interviews the brightest minds in healthcare about the transformational promise of AI and ambient intelligence for care teams.
Molly McCarthy:
Too often, technology makes caregivers' lives harder, not easier. It's time for smart technology to empower care with a more human touch. Today, I'm honored to have a clinical informatics pioneer and trailblazer, Judy Murphy, as our guest on the Smart Care Team podcast. As it relates to the intersection of nursing, informatics, and government policy, few others can match the career of Judy Murphy. After starting as a bedside nurse and nurse manager, Judy had a long career introducing and deploying an array of clinical IT solutions at Aurora Health in Wisconsin, ahead of notable roles as the CNO with HHS office of the National Coordinator for Health IT, which is where, Judy, I believe we first met, and the CNO of IBM healthcare. So we're excited to have a conversation that will draw upon the wealth of these experiences from the past and her wisdom to inform our future. Welcome, Judy, and thank you so much for joining us today.
Judy Murphy:
Thanks, Molly.
Molly McCarthy:
I'm going to go ahead and dive right in. I know our listeners are eager to hear from your vast array of experiences and really want to start out with your time at Aurora, which is now, I believe, Advocate Healthcare. You saw and really oversaw the infancy of clinical IT applications through, I would say, the awkward teenage years of Health IT, with enterprise deployments of the EMR. I know that you have extensive experience with that. And as you think about retrospectively, that journey and the evolution of Health IT to where we are today. What can you tell our listeners about what you think we got? You got right, or we got right, as well as areas for improvement or what we could have done better?
Judy Murphy:
Yeah, that's a really good question, Molly. And I would say I started in the infant years, not just the teenage years because I started doing this in the 80s. And that's important only because I think it hits on the first point that I'd like to make, and that is what was learned was the form factor itself that's used for the automation makes a huge difference. Meaning, you know, in the 80s, in the 90s, you know, we didn't have small devices, we didn't have flat screens, we didn't have good Wi-Fi. And that all played into the I'll call it the stubbornness, if you will, of the technology to actually support us as clinicians in a way that that we needed. And so when a lot of that changed, and we had mobile devices, and we had flat screens, and we had good Wi-Fi, that made a huge difference. So one of the points I want to make is the importance of actually supporting the workflow and the devices themselves moving with the clinician, particularly the nurse, and not just kind of being that fixed device. I have to say, one of the what a retrospective story. We tried to automate the recovery room documentation and there was no Wi-Fi and the devices were big. So we had this big device cart and we get chords coming out of the ceiling, and you could literally move the device like five feet this way and five feet that way because they were tethered to the ceiling.
Judy Murphy:
Suffice it to say, something like that really didn't work. I want to say it was better than nothing. But like, again, you found yourself working around the devices compared to the device actually supporting the way you were working. So really important point. You know, I think the second one is we worry a lot about data entry and the timing of data entry, but I want to also emphasize the importance of data retrieval, and that technology should be supporting that data retrieval. And we should've always done that in all the different screens and pull sheets and things that we've put into our automated documentation and even the pulling out of lab results, things, things like highlighting abnormals in a really good way and making it simple to know what lab results are, what documentation results I saw before through the use of bookmarks when bringing me in where I'd last, was looking some of the stuff that we consider just automatic today. Some of our web searches and stuff. Weren't and aren't necessarily in all of our electronic health records. So those are two lessons I think I'll just kind of throw out there.
Molly McCarthy:
Great. Well thank you. I think those are really important points, especially meeting that clinician where they're at in terms of having a device that goes along with the clinician rather than tethered to a specific point, and then the data entry obviously huge and retrieval. That's a great point. And really in today's world, presenting that information back to the clinician in a easily digestible manner. And so, just as a follow-up question to that, I would love to hear how that experience really led to some of your other positions, most notably ONC.
Judy Murphy:
So I Kinda identified the struggles that we had in those early days, and everybody was struggling. So one of the important things to do would be to share our struggles. So we weren't all learning the same lessons. So fairly early on I got involved in our organizations. Both AMIA and HIMS were professional halls for me in terms of working with clinicians and working with IT folks that, you know, were in this industry and were moving along with me. And so hearing from other people and hearing about their experiences and things was a really important point, and making sure that we were constantly sharing in our committee work, in our presentations and our writings. And we had both local groups, and we had national groups. And that really led to me meeting people from around the country who were doing this. And I fairly early realized how literally important that was, not just for the sharing purpose, but to drive the industry. And so getting involved in, you know, boards of directors and committee chairs, those kinds of things. That also led to my appointment as the national, one of the national representatives for the standards committee that OMC formed in those early days when the money started to flow related to meaningful use and helping set that criteria and things. And so I would always recommend to everybody, if you could look outside your own little world and, you know, have time to spend not just on your own job, but on sharing and networking, how important that really is going to be, not just for your career, but for your learning.
Molly McCarthy:
Now, I love that. I think I always like to say, don't reinvent the wheel. How can we share both our pitfalls as well as successes with regard to technology? And going back to that point you made earlier about supporting workflows? Think especially sharing what we're doing today and how we're really improving and evolving the nursing profession. I would love to circle back to OMC in just a minute, but I first want to talk a little bit about your time at IBM Healthcare, which received a lot of early attention around Watson AI and machine learning capabilities, which ultimately did not necessarily make the expected transformation impact transformational impact within the industry. And so, as you reflect back on that experience at IBM, would love to hear what your thoughts around AI and healthcare and what IBM Watson got right, and perhaps what they should have done differently.
Judy Murphy:
So there's been a lot of speculation about this, of course, and I would not be the one to be the expert on it. But if I look at what I know and understood about Watson, the use cases that were identified were good. They were very logical use cases, clinical trials, matching, oncology, genomics, personalized medicine those were all really good use cases. I think where there were struggles is, first of all, we know it was early and there wasn't as much, I'll call it fate, but also understanding about AI. And so people sort of thought it was like magic, you know like the computer could figure out things that an individual person couldn't figure out, which is true. But of course, the computer could only know what it's been told to know through its learning, right? And so I think we, IBM Watson, were surprised about the curation that was actually required of the data that was pulled to go into the algorithms to understand the problems that Watson was trying to solve. And when you think about things like radiology, you know, identifying tumors in the lungs or identified tumors in the breasts, it wasn't there aren't databases that have that kind of information in there. And so it was literally training, you know like Watson could say, is this a tumor? And somebody would have to say, yes, in fact it is. No, this is not yes this is. Because there weren't the large language models to actually pull from for those decision-making properties. So I think not having all of that knowledge and then the load, if you will, on clinicians for the curation of the data to make sure that it was making solid decisions, particularly in the beginning, was one of the issues. And then there was the whole trust thing. Like I said, it was early, and so folks weren't necessarily bought into this concept that they were going to believe what was actually being said or the decisions that were coming back.
Molly McCarthy:
Yeah. So trust, I think within health care, obviously so important, such a key component. And perhaps it was just before its prime, so to speak. And speaking of which, in terms of hype and magic, we've heard a ton this year around ChatGPT AI, Generative AI, and I think every vendor and startup is now suddenly making claims around AI and health. And really, I think as a health system leader, how do you separate kind of the hype and the noise from—the truth?
Judy Murphy:
So I think we have to first of all learn. We have to understand what AI can do and what it can't do, and understanding how it works and what data was actually used to make the decisions that is coming forward and giving to you. And a lot of the tools that I've been hearing about being developed, you can, you know, I'll call it click a button, but it could be some other way of getting to it. But that once you get something back from an AI tool, that you always have the capability of clicking something to see what knowledge was used to make that particular decision. I will say we have been in the healthcare industry very, very, very EHR-centric. And because of that and because of what we talked about minutes ago related to the workflow of clinicians, it is not going to be simple to use a tool that is not in the EHR, because you'd have to break your workflow, turn to something else, or click on something else, or go to another app and use that. And so, in the context of providing patient care, I think the implementation of something like chatGPT is going to be difficult. Now that being said, if I'm looking for information to help make a diagnosis on something rare that I don't understand, or if I'm looking for predictive analytics, these now would be really good ways of using those kinds of tools outside of the actual delivery of patient care. You know, when I've got the time to do those kinds of things, and I see that's going to be so in my mind, sort of the early things we will use it for.
Molly McCarthy:
Yeah, I think you're 100% right in terms of EHR-centric. I've spoken to people who have mentioned, and I'm sure you have too, that if it doesn't integrate with the EHR, then it's a no-go just because of the amount invested. So I think that's a really important point. And the other point is going back to talking about the workflows. You don't want to interrupt a well-established and well-documented workflow. So I think that's a really important point. And one thing as we think about the emerging use cases, you started to talk a little bit about that, but where do you think AI can make the biggest impact for care delivery and for patients? And then any thoughts around how we can utilize it to make the life within the hospital for nurses and other caregivers easier? So it's not encumbered cumbering them.
Judy Murphy:
So in my mind, the one of the best use cases is helping us evolve evidence-based practice and helping us really understand, as nurses, what are the things that we do that make a difference to the patient's outcomes, and what are the things we do that we just always sort of done and could possibly, could possibly drop? And so with the electronic health record being implemented now for many years, we've got data that we didn't have 20 years ago or even 10 years ago. And so looking at using AI to forage back, if you will, into the care that we provided and helping us understand what things really, really do make a difference to patient care, both in the inpatient setting and the outpatient setting. And how does it change outcomes? You know, one of the things that when you think about traditional research where you have to hypothesize and then we test that hypothesis. Right. And one of the beauties of predictive analytics using AI is that you don't start with the hypothesis. You're looking for the similarities; you're looking for the trends. And so maybe some of the things that really make a difference aren't things we've looked for, they're not things we've hypothesized. And so in my mind, that's what's really going to help us advance evidence-based practice.
Judy Murphy:
The second area is in telehealth and home monitoring. If we want to do something like home monitoring broad scale, we're going to have to have a way of collecting information and then sifting through it and knowing which patients to have real in-person touchpoints with, right? And he can help with that, or it can look at the trends of activity or the trends of intake and output, or the trends of blood pressure, the trends of just about anything. And if something gets outside that individual patient's norm, you know, alerting the nurse so she knows that's somebody that she has to call or has to visit, kind of a touchpoint. So that's the true expansion, if you will, and the capabilities of value-based care and population health management, where, you know, on a broad scale now we're looking at lots of patients, and AI can help us know which ones need that that touch point today which won't see them tomorrow, which ones can wait? That type of thing.
Molly McCarthy:
Great, so one I heard from you was just evolving evidence based practice. And I think the second one, I've spent a lot of time in this area just thinking about home monitoring and being more efficient in how we care for our patients across the care continuum. So those are two great points. And you mentioned the word sift through the data. And I think that's critical that as we have you mentioned we have years and years of data, are we using it to improve care? We need to delve back through it with AI and then quite frankly, present it in a manner which is digestible by our clinicians. I want to move on, and I want to talk a little bit more about your time at ONC. And as we think about what we learned from our EMR experiences regarding government involvement, how do you think about the needs for oversight, standards, and regulation today related to how we're applying AI in healthcare? I know there's been a lot of talk about that. I'm involved with some work around that as well. So that's my first question. So I'll let you answer that.
Judy Murphy:
Sometimes we need government prodding, I'll call it to get things right. And, you know, think back on the stimulus package where we had the meaningful use program. And the one thing that sticks in my head that I always wonder about is patient portals. What patient portals naturally have evolved if they weren't mandated in the meaningful use to criteria? So curious question. My guess is eventually, but think it would have taken a lot longer. And I say that because the initial criteria was you had to have a portal and one patient had to use it. I mean, it was like a nothing criteria. And we sit here today and we know just about everybody uses the portal. And the percentages in general are quite high to look at your lab results, you look at X-ray results to see your physician's notes, to see your after-visit summaries, all those kinds of things. And so that's a stimulus where there was involvement, and it probably was needed. You know, another really good example is standards. I don't think our vocabulary standards, I don't think our transmission standards for interoperability would have moved along. The governance would constantly been prodding it and everybody watching what's going to go in the USCVI next, you know, kind of thing. And I feel like that kind of, and I keep calling it prodding because they don't do everything, but they do little pieces that kind of spur the industry to start thinking about some things that they should be doing. That being said, I do not think everything should be regulated just like today. It's not, you know, regulated through the government. I mean, maybe some of the ways we're going to get the kind of standardization and evidence-based practice that we need are going to be our other regulatory bodies like Joint Commission or like Fraschini, you know, kinds of things where a standard is developed through consensus if you will, and wanting to be able to get certified as compared to the government actually mandating it.
Molly McCarthy:
Those are great points. I love your comment about spurring or, you know, using the government to boost and really push along some of the different models over time. Patient portals, for example. It's a great comment. So without slowing innovation, how can we ensure AI-based solutions launched in market actually really improve quality, safety, equity, and clinical outcomes while lowering costs at the same time that quintuple aim? The other piece that I'm going to add in there is not just clinical outcomes, but also improve, as we talked about earlier, the lives of our clinicians.
Judy Murphy:
Before we launch these applications. I don't think we can ensure anything. I'd love to be able to say that, oh yeah, they have to be perfect before anybody uses them. And that's just not going to be the case. So I think there is going to be some need if you will, that as these kinds of products are launched and used, that there's people monitoring and watching and testing and validating that. In fact, the tools are giving the kinds of answers and solutions that are in fact appropriate. This is going to be an evolving thing for a really long time. And I hate to say we're going to learn as we go, but I do think that's what it's going to be. We're going to learn as we go. When we think back on our EHR journeys, we got some things right, we got some things wrong. And so we stand here today, and we're fixing the things that we didn't get right, and we're getting things better. Are they going to be perfect? We don't know until we try them, right? Because we don't have a good lab where we can test these kinds of solutions until they're out in practical use. And we're getting feedback from clinicians who are doing practical things with patients every day. Now, again, we have to have some guide rules, right? We can't just go blanket and say everybody can do everything. And so maybe that's where the regulatory agencies and or maybe even the government is going to come into play where we create just some think of it as the bumpers on the side of a bowling alley lane, right? You know, where you can't go in the gutter, it might not be a strike, right? But you've got to prevent people from going into the gutters. And there are probably some overall big rules that are going to be required related to things like ethics and safety that will be required. But otherwise I think we're going to learn as we go.
Molly McCarthy:
So great points. And I think the second point you made that I really want to hone in on is feedback from clinicians, that's critical at all points in that development process. So I just have one closing question for you. So most of our listeners, health systems, CNOs, CNIOs, and their teams, and obviously you've got a really unique lens, having walked many miles in different shoes from bedside nurse to CNO of IBM Healthcare. And I really want you to leave our listeners today with some important practical piece of advice right now with keeping in mind the healthcare market and arena.
Judy Murphy:
That's an excellent one that I have thought about many times over my career and written some articles on. In fact, the title of one was The Best IT Project is not an IT project, and I hone in on that because technology advances and then we see its advancement and we figure out how to apply it to our work. And that's backward. In reality, we should be identifying the practice change that we want to implement. So we should be visualizing how we want care to work, and then looking for tools that are going to change toward that direction that we've identified. That doesn't always work, right? Because I'd love to say we consider how we want to change practice, and then we go to a vendor and say, we need you to build X, Y, and Z. That just isn't how the industry works and never has been. But as close as we can get to that and to take these sorts of implementations, especially when we start talking about AI and nursing and thinking about how the practice is going to change and how we're going to make or use that practice change to make a difference in patient care and patient outcomes has to be front and center as compared to the trials and tribulations of the actual implementation. You know, the implementation is so hard. We get so involved in math that we don't think about that practice change and how it's actually changing the way we work and changing the way we deliver patient care. And so keeping that change front and center and letting some of that other stuff kind of become peripheral is probably the advice I'd give very practically to all those folks who are looking at doing implementations in the next five years.
Molly McCarthy:
Well, thank you, Judy. And listeners, you heard it from her right here. The best IT project is not an IT project, I love that. And to your point about really identifying what needs to change within the nursing practice or the clinical practice to improve patient outcomes is critical. So thank you so much for being with us today, Judy. Appreciate your time.
Intro/Outro:
Thanks for listening to the Smart Care Team Spotlight for best practices in AI and Ambient Intelligence, and ways your organization can help lead the era of smart care teams. Visit us at virtualnursing.com and for information on the leading smart care facility platform, visit care.ai.
Sonix has many features that you'd love including automatic transcription software, collaboration tools, transcribe multiple languages, secure transcription and file storage, and easily transcribe your Zoom meetings. Try Sonix for free today.
"One of the beauties of predictive analytics using AI is that you don't start with the hypothesis. You're looking for the similarities; you're looking for the trends. And so maybe some of the things that really make a difference aren't things we've looked for, they're not things we've hypothesized. And so in my mind, that's what's really going to help us advance evidence-based practice." - Judy Murphy
care.ai is the artificial intelligence company redefining how care is delivered with its Smart Care Facility Platform and Always-aware Ambient Intelligent Sensors. care.ai’s solutions transform physical spaces into self-aware smart care environments to autonomously enhance and optimize clinical and operational workflows, delivering a transformative approach to virtual care models, including Virtual Nursing.