Good afternoon, everyone. Welcome. Thank you for being here with us today on this Thursday afternoon. I hope you're enjoying your time here. I hope you're in the right place. This is the CAHPS PCI Registry track, just to make sure. And this topic is around easing the burden of data abstraction. Now let me just say that is incorrectly titled because it should never be called a burden. So it should be titled easing the effort, right? Because everything of value takes effort, right? So this discussion, we have some very distinguished guests here that will walk us through some approaches to easing the data abstraction effort. And our first speaker, Serena Felcher, she'll be diving into a little bit of the AI approaches to data abstraction. And then Lisa Foster will be diving into some more of the manual approaches to creating efficiencies with your data abstraction. So without further ado, we are going to go ahead and start with Serena. Thank you very much. She said distinguished, and yet here I have my flip-flops on. Thank you, Florida, for the weather and for me fooling outside very much. I'm very happy that I have toenails that are done. So good afternoon, everyone. I hope you're having a wonderful summit. Every year for the last almost two decades, I was sitting right where you were sitting. Coming here every year, listening to those who are at the podium with rapt attention, taking in their wisdom, taking in their innovations, their tricks and tips, and taking them back to my own organization and kind of paving the way through that. And so I'm very happy to be here at the podium. I hope my presentation illuminates something for each of you and you take back to your own organization and it starts a dialogue. So good afternoon. I have nothing to disclose other than the fair use notice and disclaimer. So everything that's on my slides, you'll see I have a little blurb on the bottom if it was taken from another source. Learning objectives. So we're going to discuss today some trials and tribulations of implementing AI technology and how it was implemented at our hospital and our organization, exploring the data registry role in supporting and driving quality with assistance of AI, identifying time-saving techniques for improving that. And then lastly, where is AI going for the future? So this is my hospital, Sutter Medical Center in Sacramento. And as of 2015, we were two separate campuses, kind of distinct campuses, and they merged together into one campus, one huge building. And then from 2015 on, we've been building up that campus. And so most of the buildings are new and we're just growing. Sutter Medical Center is in Sacramento. As you can see, we're in Northern California, straight in the middle. We serve a population of one and a half million people in the most populous state in the United States, which is California, of almost 40 million. Now my accent will prove to you that I'm not from California. I'm originally from New York. So those from New York went to college in Charleston, South Carolina, love Charleston, and then found myself transported to California. So when you look at Sutter Health and our network, we're a community-based, not-for-profit organization. We serve the greater area of Northern California, three-plus-million patients, large area. So if you look to your west, we occupy San Francisco, all the way to our east to the Sierra Nevada Mountains, almost to the border of Nevada. To our north, we have the wine country, which I like. And then all the way to the south, back to the beach, Santa Cruz. So it's a large area. Three million patients are being seen by our network. Oh, I'm going to go back a little bit. We have 24 hospitals, 36 surgery centers, and very proud to be one of 14,000 nurses. I am a nurse for 25-plus years, 53,000 employees. And so it's a very large organization. Sutter Medical Center itself, I'm very happy to be part of an organization that thrives and that really pushes itself to the limit to make sure that it's meeting quality and patient experience. We're named a best hospital in California, top 100 hospital in the nation. We have many accolades to our name, but I think the two that really stand out for me, and I think you all will agree, is the first one is we got comprehensive cardiac certification. Second only hospital west of the Mississippi. And the only other hospital would be UCSF, and they were the first hospital in the nation. So very proud of that designation. We also meet, we're a three-star TAVR program, a four-star CABG program. And so we go on and on, but it all starts here at ACC, coming to these meetings, diligently learning from our peers, diligently learning from those who were on stage, taking that back to our organization, and really kind of paving the pathway forward. So I'm very happy that as an organization, we took a journey to the future, all aboard. And I use the train as a kind of analogy as moving forward, and that's what we're doing at Sutter Health. We're moving forward. And before we moved forward, we had the pleasure of getting a new CEO in 2022, our new CEO, Thomas Warner. We kind of took him away from Ochsner, which is in, I believe, New Orleans. And he did some wonderful things there. He is a true believer in the digital health. He's a former, like I said, CEO of Ochsner. And Ochsner was named, year after year, the digital-wired health care system, the most digitally-wired health care system in the United States. So taking that, he pioneered it with partnership with Apple and Microsoft, and we're all familiar with those companies. And he embraced digital health innovation, deploying digital health solutions, and tapping into the true growth of what digital can enable a hospital system to do. Just in a short period of time, with his vision and the growth in digital at Sutter Medical Center and Sutter Health in general, one-third of all of our hospital appointments are made online digitally. We have digital tools for our patients, mental health, parents, caregivers, learning, education. So it's a real partnership for the clinicians and the digital world itself to come together to optimize where we go forward. And so I believe that is where we're heading to the future, all of us. Health care is one industry that needs constant, present, and future comprehensive innovation. Eccentric, always, with efficiencies, with personalization, guaranteeing our patients the best health care outcomes available. So how did we do that? Well, it wasn't easy. It truly wasn't. We started this journey more than two years ago. And as you can see with my slides, we had 28 sites, four distinctly different CBIS systems. So many of you guys, if you come from a big system, you can relate to this, that it's, you know, diverse areas had a diverse way of doing things. We had structural reporting that was also quite diverse across our system, seven different systems for EKG, structure reporting. And we had to find a way to make this a seamless one-system approach, a Sutter health approach. And so where are we two years later? Well, we are now a 28-site with one CBIS system. We have invasive structural reporting, non-invasive structural reporting, EKG reporting, all one type. And then the most important thing that we brought to the table, or that was brought to the table was an implementation of an EHR-based data registry repository. And why is that important? Well, to paraphrase what was put in the American Journal of Medicine, to survive in today's competitive landscape with limited budget, health systems must maximize every resource. Learn how to effectively, clinical registry tools can save the organization money and discover how augmented intelligence, the merging of human expertise and artificial intelligence holds the most promise as we advance into the future. So I'm going to give you kind of a lesson here a little bit. As a nurse, if you ask me what to do for a patient, very easy, but I'm not an IT expert. And so looking at augmented intelligence versus artificial intelligence. Artificial intelligence is the ability for the computer to think for itself, to come up with a solution. Artificial intelligence is taking my brain power with the computer's ability to automate things and now coming up with the best optimized solution. So from here on end, all of you have the ability to do this at your own organization. As you can see, the top six EHR softwares, which compromise 95% of all U.S. beds, somewhere in here all of us are using one of these systems. They all have the ability to do what I'm going to describe to you in the next few slides. They have it. It's the implementation of it that sometimes scares administration and hospital systems. But I'm very grateful that Sutter took that leap of faith. It's been nothing but a positive move forward for our organization. Sorry, give me a second. So I'm going to again, not an IT expert, but I'm going to give you a little bit of IT. How does data flow? How does it get where it gets? Even at your own organization currently, this is how it really is working. You have your physician in his office, in the hospital, a nurse practitioner, however it is that you're documenting, it's an EHR, 95% of us are using an EHR. It goes into the healthcare record and then it goes into the data warehouse or the ETL as well. So this is really combining all kinds of sources of data or information from various sources. It could be your cath lab. It could be your echo. It could be your doctor's office. It could be ICU. It's housed in a data warehouse. Application, I'm going to just switch over and I'm going to take you to that arrow that says application server. Application server is another fancier way of saying our EHR data registry repository. So you can take it straight from the data warehouse itself, certain things, and just push it straight into the data registry repository. That's one way you can do it. Or it can move through the system. Curated data is really data that's looking at groups of areas. Your ER admissions, your congestive heart failure patients, data sets, things that you want to then take and pull using for either research or other areas that you need for it. And as it goes up, then you get to API, API being a program. And it can be whatever your organization decides to use. It really is an interface, telling two different programs to talk to each other in the same language. And then it can also go up to the application server. And the application server being the data repository. I'm going to toggle back and forth for a second so we have a clear understanding. So at Sutter Medical Center we are on an EHR system as well. So our data, going into our data repository has two ways of getting in there. It can come from the data warehouse itself, as you can see, which is EHR and non-EHR data. It can go into the Chronicles, which is real-time data happening at that moment and being transposed straight into the data repository. Or system analytic database, which is large amounts of long-term looking at a patient maybe who has been there for 20-plus years. And as that gets uploaded and moved, it can be moved into the data repository. So what is the advantage of moving it from where we were currently, which was a third-party vendor and or using ACC, and I'm going to go through how that has made it much better for our system. So we're harnessing now and exploring the potential of AI. We are now in our EHR system. The data is all there. Instead of transposing it now to a third party, it's staying within our own system. It's moving to a data repository within EHR itself. Again, the train showing the progression forward, because we're all under the same challenges, making sure that we meet governmental, public, health, payers, regulators. And so we need the data, and we need it to be efficient, and we need it to be data that we can trust, in a system that we can trust. One of the advantages of having it in an EHR is it's now accessible system-wide. I'm not putting it into a third party or ACC, and somebody across the state can't access that data. It's in our EHR. And so it's accessible to anyone across our system, with the correct permission, of course. The ability to onboard new data registries to meet the system needs, to do the work in one place, which will increase efficiency, simplicity, and decrease the possibility of errors. And luckily, it's very much easily integrated with ACC and most of the other data registries that we employ at our facility. We have the ability to transport it externally with no issues, and the ability to capture discrete fields, if needed, for clinical research or other areas. So how did we do it? Well, we did it painstakingly. But the outcome, the outcome after it all was said and done was a system that was intuitive. So let me give you an example. We do about 11,000 cases in three registries, and those three registries, we're going to just use this as an example, our AFib Registry, our CAHPS PCI Registry, and our ICD Registry. When you combine the number of data elements that we're all collecting, that's about 900. We do far more than that. We do more than 15,000. But we started off with the first three, and the first three being CAHPS PCI, again, AFib, and ICD, with the idea that this was going to be deployed across multiple other registries. So when you look at 11,000, so let's take just CAHPS PCI. I can have AI, automated intelligence, which is transposing data that's already in our system and moving it into the data repository seamlessly without a data registry person even looking at the data elements. I can do that for 9,000 cases and about 43 data elements itself. Well, take that. How long does it take a data person to abstract those 43 data elements? Let's say they're really good and they do it in six seconds. So I did a timing. Six seconds times how many 43 data elements times 9,000 cases, and we get 2.3 million seconds or 38,000 minutes. AI is doing it without a blink of an eye, and they're doing it and saving 645 man-hours. That's tremendous, and that's just in our pilot program. That's three registries. We haven't even thought about the other 15-plus registries that we can now deploy this to. And so now we're saving 645 man hours that can be used for process improvement, steering and moving the pendulum to real changes, not looking at the data. So I'm going to show you how it's done. And this is just an example of the 43. Real stuff, the patient's last name, their race, their ethnicity, their age, their weight. None of that needed to have my eyes look at it. It's pulled from a data warehouse and it's plugged into the registry waiting for that data abstractor now to continue on to get the rest of the materials. But we further made it even better. And how did we do that? We looked at how we can use AI to pull more data, data that normally was typically done by the data abstractors themselves. And this is really a synergy between IT and the data abstractors, hypertension, things that require a more pronounced or a more thought process that AI couldn't do. How do we program, how do we create an algorithm to allow AI to be able to pull these data elements into our repository without a data abstractor actually doing it? And so this was a real partnership, and I have my beautiful Sutter people in front. Some of them were instrumental in that, actually pushing this through. It's a synergy with IT. They speak one language, I'm a data abstractor, I'm a nurse, I speak another language, merging our two languages so we understand one another to get what we needed. And then creating an algorithm that can do it. And so, so far, just in six months, we're able to very shortly get some of these also data elements pulled through an AI process. And so it's quite encouraging for what we would like to see for the future. So still got some time on my clock. So let's look to the future. What's the next step? Well, the next step for Sutter is deploying this to other registries, taking advantage of AI. 645 man hours is going to be 10 times that much in the next few years as we deploy other registries. But what is the real future? How do we really optimize AI, augmented intelligence, to work for us? And so I'm going to tell you. Structural reporting, it's the nirvana. It is the future. It's using AI and registries together. It's working data integration into the workflow itself, not separate from the workflow. That means in your EHR itself. It's having the information provided by the physician in the EHR and then transposing it straight into the data repository. It's not an easy task. It's changing an ancient paradigm, a physician model that normally they're pushing out their documentation by free text, dictation, or other antiquated venues, and then transposing that. We, the data registry people, we transpose what they've written now into a data repository. We're changing that thought process to let's do it. Let's do it once. Collect it once, and now we can utilize what we've collected for data registry, for research, for really patient population changes, et cetera. So structural reporting collects all the information needed one time, and then again allows that data to be used multiple different ways and for multiple different purposes. I'm going to show you in the next slide what it actually looks like. It's a marriage of the cath lab processes and data collection at all levels, the hemodynamics, the medications, the devices we use, what the physician portion, all digitally being collected and then being uploaded into the data repository itself. Structural reports, data, not words. So physicians, you know, they write this long procedure or op note. There is a better way of doing it, creating the data that's needed for what they've done and populating what they've done now into a report to reflect what has happened, incorporating this vital data element in a drop-down system that creates a summative assessment. And then taking that again and using it in however you really want to use it, a repository, research, or process improvement, quality improvement, looking at your patient's demographics and other things. It really helps speed up efficiency, effectiveness, quality, production, because it's all within one system. My eyes are looking at it one time. It's looking at it in one system, not multiple systems, not a third party, and then moving over back and forth. I'm looking at it in one system, and it's being transposed automatically into the data repository itself. So how is that going to look if we did move to the future? Structural reporting really is going to look and mirror somewhat what the ACC, some of the data elements, we've seen it in practice. We've seen the demos. It really is going to look what the ACC, but it's also going to incorporate some other things. It's a collaboration at all levels. So as you can see on the right-hand side, it's what the CV tech is doing. That's going to be pulled into that report. It's what the physician is, besides writing, he's going to click mark. I did a left heart cath. I did this. And it's going to now summarize it into a report. But it's also going to automatically, the device is used, it will be a scanned situation. You scan it, and it's automatically, so the doctor doesn't have to remember after, did I use a 2.0? Did I use a 2.5? He doesn't have to remember. It's already now in his report because it's been scanned. It's digitally in the report. So he can review that report as he's making his comments and his notes. Now it's not to say that he doesn't have the opportunity to still make free notes. Of course he does. But it really kind of shortens his time in dictation. And because it's seamless, it's immediate. It's not waiting a day or two for him to dictate. Some of it is being done at the bedside, in the cath lab, click, click, click. And we're getting already a summarized assessment of what happened in the cath lab. And now here's the flow. Flow of data comes into play. All of that is being uploaded straight into our data repository, which is in the same system, our EHR. And so it's being uploaded. I don't have to look at my right heart cath, my wedge pressures, let's say if it happens to be for TAVR or something. It's automatically going to be transposed through AI. I don't have to look at what medications were given during the procedure. It's uploaded. If it's needed for that registry. And so it's really an innovative way of moving things forward. Taking away from what we do, which is very manual and automating some of our... And one of the things a lot of... I work. I do data abstraction as well. So you say, well, does that mean my job is going to be lost? It just means our focus is changing from the data abstraction to true looking at process improvement, quality improvement, looking at how we can make our hospitals the best that they can be. The industry itself is a tight industry. It's competitive. Everyone wants to be the best of the best. And how do you become the best of the best? Accreditations prove that you are the top notch. We're all vying for the almighty dollar. And how do I compete to get that almighty dollar? And it's being the best of the best. And so in order to do that, collecting the data is extremely important. But if the data can be collected seamlessly through an AI process, and my intentions can be then to further move that pendulum from my program, move that pendulum to true process improvement, quality improvement, and become the best of the best, well, then I've done a very good service for my organization. So that being said, give me a second to think. That being said, that's it. I truly hope you enjoyed it. I have another presenter, but I hope you all have questions. I do. Because, like I said, I've been in the audience 18 years, and I've learned so much from those who were up here. And so if I can impart some of my wisdom to any of you here, well, then I've done my job. So have a great day. Lisa. Thank you. All righty. I feel like it's like point counterpoint, you know, because now I'm going to go old school and we're going to talk about all that we're doing manually and how to make our abstraction the best we can. All right. So our objectives for today, I want to discuss how first off to set up for success. And then I want to talk about how to leverage our EMR, custom reports, patient identification for our list that we have to abstract off of. And then also discuss the value in that term I think we've heard a few times, submit early, submit often. A little bit of background first. To start off, I'm from Baylor Scott & White Health System. Our health system became Texas' largest health system in 2007 when the two bigs merged together. I then, and you can see here the number of facilities we have, the number of patients the health system touches. Taking it a step further, I am with Baylor Scott & White, the Heart Hospital in Plano, Texas, just north of Dallas. We have Heart Hospital Plano, McKinney, and now Denton also. And I also help with the MSAs or managed agreements with two of our other Baylor facilities. I did want to highlight here we recently had our STARS again awarded to us for both STS and through VQI. And now that VQI touches NCDR also, I wanted to add that to my slide here. We participate in our department in, I think, counting the different modules, over 17 registry types in total. Here's our volumes for Fiscal Year 23 just to give you an idea of what our Heart Hospitals are doing. And then just a little more about me specifically. I am the CAF PCI RSM for the facilities I've mentioned. I have abstracted over 10,000 cases now. They had a great celebration for my 10K. I tell everybody, does that mean I'm dedicated to my job or does it mean I'm old and I've just done this a long time? I don't know. I have a great cath lab background. It's been my joy since I came out of rad tech school. I went right from being an x-ray tech into a cath lab. And I haven't really left that since then. And I now am in a great quality department. We've gone from two staff members when the Heart Hospital opened, now up to 13, 14. And we're growing daily. And just again, let's get back to it. Why are we here? Quickly look at my objectives and let's move forward. So starting off, we want to set up for success. When our newbies come in, I think I've heard from several of the presentations in the last two days the past experiences where you were just given a stack of papers or a dictionary, go learn it, figure it out. But at the same time, let's abstract it. But now we want to set them up right from the beginning. And then along with that, the next step is case finding, what kind of custom reports we can use to make that easier for us. And with that great EMR, how do I find what I need? Prior to this EMR system, we had limited data within that medical record. Might have been on paper, some electronic, but it was limited. Now the EMR, it is so vast. And you can get caught up in spending too much time sometimes trying to dig for that piece of information because we are all these type A personalities that we know we can find it. So back to setting up for success. It starts with our onboarding process. It's not just a hot body in that role to do the abstraction. In order to have a successful abstractor, we want to start from day one with them, making sure that they have access to everything they need first off, making sure that once they've finished their facility-required education pieces, then we move them to the abstraction piece. And I'll touch base a little more on my next slide about what NCDR offers us. But within our facility, we've created an onboarding system for all of our new RSMs or abstractors that come in for all of the registries. We have a template made, but then we adapt it for each individual and each registry. And first and foremost, we want to get them the right resources first, give them the knowledge before we put them into abstracting. And then being realistic with our expectations. Everybody does learn at a different pace, a different style. You don't know what might be happening in your organization during that first month. You have this new member where your attention might have been pulled away. So being realistic and being adaptable. Again, back to NCDR. The first step for us is getting them back to the modules that are offered. Start Here is truly the Start Here place within the NCDR site. And we do build this into our orientation program. We have them complete all the modules offered before we even start with abstracting and even looking at our data collection forms. And even though some of those modules are a little more advanced, we still have them do them all in the beginning. And then as we go further, we can come back, take them again, kind of when those pieces start to come together. Once we've done all the modules and we're starting to look at actual abstracting, we've created a quick start guide. And I'll take a step back. We do use a third party that we abstract through. And then it goes on to NCDR, of course. So these slides here, you'll kind of see us referencing that. We've created this guide that walks the new person through step A all the way through the abstraction. And it's something they can have next to them in the beginning, in the early days, and flip through. And then we include a lot of the NCDR resources within this guide that we created also. And again, as I mentioned, we make this onboarding packet that we can adjust per person. We always have kind of our target model where we think they're going to be within set time frames. But then based on what we're seeing in their audits and how their knowledge is, sometimes we have to expand it a little longer. And then there's been instances where that person ramped up quickly. We were able to move them forward with expectations on productivity sooner than expected. But again, with that, it's kind of looking for the sweet spot. I mean, you can have someone that is abstracting 10 to 12 cases a day right off the get-go, but their accuracy on the IRR audits are up and down across the board. Then you may have someone that's only abstracting a couple of cases a day, but their accuracy is awesome, which you want the sweet spot. You want the combination between those two. And we referenced some data that NCDR had shared with us several years ago. And that gave us our benchmarks that we compare our IRR audits to for each new abstractor. And we have expectations as we're doing those audits that they have to meet certain benchmarks before they're able to proceed on to the next step. And then another piece that we include is our time studies. We recommend in our orientation, we'll do that at 90 days. And then we recommend doing it again at six months and then annually. I myself like to do one every year on myself. Two things, it lets us see how long it's taking us to abstract, which is important. But then the way we have it set up, we know that a lot of the time we don't abstract one case in one sitting. We have other things we're doing in our day. So this also helps us to identify the amount of time we're not abstracting and what are other priorities are that are competing, which then you can also link into your productivity, which links back to your budget. And fiscally it may help to support the need for more staff. Next is really knowing your definitions. Like I said, we used to, you know, day one you were delivered a stack of papers and you read through them. We're getting a little more techie. It's more online now. But our data dictionary is still our Bible. It's where we have to go. We need to know the definition to make sure what we're putting in is correct. Our RSM calls, our case scenarios, our FAQs, and the great Contact Us on the NCDR site, all of those are great resources. And what we've done is we've taken and we've created a OneNote for all of this. So every time one of us reaches out to you in CDR and asks a question, you email it back, we dump it all into this OneNote. All of us PCI abstractors have access to this, and it's our shared resource. We add to it all the time. Next big step for everyone, especially our newbies, is case-finding. And I don't know, I only know my experience on how we case-find. I would assume that everyone, it's kind of a task. You want to make sure you have that right list of patients that you are putting into your data. We work with our EMR. We've created some custom reports, which I'll go into in the next few slides, and then working closely with our cath lab. I'm in contact every day with our charge person in the cath labs. We talk back and forth. Sometimes she might find a case that I miss. Sometimes I find a case that shouldn't be on my list that the staff charged a wrong charge. And being proactive. We want to stay on top of our case list daily so that we know what our workload is, how we can better align our time for that day of the week, other priorities coming up. So what we did is, working with our EMR, and I'm sure many of you have already done this, but we created custom reports based around the procedures we were looking for, and then also per facility. And each morning, that's the first thing I do. I run the prior day's list, send it out to some key individuals that also use that list to make sure cardiac rehab documentation was completed. They use it to double-check discharge medications for PCIs to make sure they caught the right medications. But then I use this to anchor my cases. So every morning then I anchor these, basically making a placeholder. We have that list there. We know what we need to work on today. And the reports are based off of procedure charges. I think we have, I think there's like 27 procedures behind the scenes, you know, between your stents, your arthrectomy, your mechanical thrombectomies to capture the PCI. And then, you know, prior to the EMR we're on now, we used our cath labs data management systems, or their CVIS or hemodynamic system to pull the same type of list daily. So I know that not everyone has access to the latest in EMRs, but you can still get lists out of your cath lab systems. Is anyone still using paper forms that you used? I mean, we used to take them at my other facility and basically give the cardiologist the NCDR data collection form and he got to fill it out. We would put a patient sticker on the top and here you go. For our facility, everything for the PCR registry is 100% chart abstraction. We do not use any forms. Some of our other registries we participate in they are still using a limited form, but trying to get away from that as much as possible. Number one, that was a huge physician dissatisfier filling out all that paperwork. And again, I mentioned it earlier, now with our EMR, there's so much data available. And then digging through it sometimes. But again, work with your EMR informatics, your IT people to make sure that you're getting the right information and your IT people to make custom reports a step further. So what we've done is taken and we've come up with reports for each registry that's kind of our starting point. So we have one for PCI, STS has one, I think LAO, EP. And it's once we have our work list for our patients that we're abstracting, as you click on that patient it immediately takes you to this summary report. And we tried to mirror it as best we could to follow the flow of how we're abstracting our cases. It's a starting point though. I always, this is where I feel like it's the point counterpoint. I would love to trust this 100% and believe in it, but I'm not there yet. And I know our definitions sometimes are so specific. I don't know if our EMR is quite to that level yet to knowing that that is the lowest hemoglobin before we went back to the cath lab for that emergency PCI after the initial one. Excuse me. So I do use this as a starting point. It's a great place to give me lots of tips and tricks and things that I'm kind of noting in my head to look for as I go a little further. But then there's other pieces in here that I do use. You know, it's a quick place to start seeing that HMPA information. And it's long, it really is, but it really truly follows our data collection form. And then my next best friend within the EMR is that search box. You know, that I think saves me the most amount of time is just, you know, plugging in NYHA and amazingly, it'll come up. It's, you know, the doctors tend to sometimes forget some of the key things for us in their HMP, but you can usually find it. And when we do find things in unusual locations, we make sure that we note it when we're filling in our abstractions so if by chance we should get audited several years later, it says right there. You know, I found it back on the 2003 note that yes, that patient's mom had a heart attack at 42. And again, leveraging your EMR, build what you need. It's gonna save you time in the long run. We had a mortality project several years ago and we built this scoring tool that's now completed on every inpatient before coming to the cath lab that gives us an idea on their potential risk of mortality. And then over a certain level, it requires a second opinion. We then took it a step farther and we did this with the AKI tool as well as bleeding tools. And I know all of this is available through apps, especially on our ACC site. And the doctors do use the app, but they wanted something that was also in that patient's record documented. So they're filling this out on every case before coming to cath lab. And then it also helps me. Sometimes they'll mark something on here that I may not have seen early on, but that's my clue. Hey, he did mark CHF here. I've got to go back. It must be documented somewhere else. And then in the future, well actually it's now, we are creating an AUC tool that will be part of our record. It's gonna go into part of our structured report. It'll be one of the additional tab. It's created. We're just in the build stage within our EMR right now. And again, submit early, submit often. What is the value in that? First of all, it helps us with our weekly productivity reports. We do these based on number of cases, the length of time it takes us to abstract and finalize them, or turnaround times. And this data has been used to not only show us where we were, where we want to go, how many days we want to be within, but again, it has helped support growth in our department and it proved that we needed additional support in abstraction. And then it also, for me, the number one reason we're so real time and we do this submit early, submit often, it makes it so that we can real time develop process improvement, identify needs that we can quickly address. Just a simple example here is with our discharge medications we are typically about 99% on our outcomes for these, but a big piece of that is because we are real time and that next morning I'm already looking at who discharged last night or yesterday and reaching out if we see a missed med. And hopefully it's just a piece of missing documentation. It wasn't truly a missed medication. And then when we are submitting often, we submit at least every other week and then the following Monday we're going into our dashboard, reviewing that submission, looking for what we need to go back and look at. And here's just a quick example of two different facilities. Hospital A was submitting often and you can see they have green scores each time. By the time they got to the end of their quarter and ready for final quarter submission, they had already reviewed all of that data. They didn't have to spend days looking at a quarter's worth of data. And then we have another facility that was not submitting as often and then they were also just not having the time maybe to really dig deep so you can see all the yellows. So that meant when it got to their quarter they had to spend a lot of time and by then you're also past the date that you can make any chart adjudications or amendments as needed. And when we do our weekly submission in the following week, as I go back into my dashboard, do my patient detail, get my list, export it out to Excel, and I start to work through that, I keep that list so that as I'm going through the quarter I can just add on to that list so I'm not redoing what I already did. It's keeping a note for myself. So by the time the quarter is done, typically I'm final submitted within a couple weeks after the end of quarter. So wrapping up, I hope that after listening to this you're able to hopefully help set up for success, kind of give you some ideas on how to use your EMR, how to come up with some custom reports and learn the value in submitting early and often. And I want to thank Christy Verschelden from my department. As I was creating this, I realized almost every resource we use within our department, Christy created. And she's the one that would never take credit for this, but I wanted to really acknowledge her. And she continues to build these things that when I saw the potential new dashboard yesterday with Power BI, I'm like, that's Christy. So I was very excited. Just a snapshot of my department. I'm really proud to be with this group. And then my favorite committee that I'm on is our Positivity Therapy Committee with our two therapy dogs. And just on my resources again, I wanted to reference Christy and then of course our outcomes. And I'll go back. I hope you have any questions for either of us. Thank you. I'm glad. Lisa, come stand with me. We'll start with the leader. Is that me breathing? It's you. We're all hearing it too. Oh my gosh, that's weird. I know. It's the Florida. Kudos to both of you for doing such a great job. Thank you. I did wanna point out Lisa and Christy have been great contributors to the participant resource sharing pages that we have. So she didn't give herself kudos for that, but you guys have done a fantastic job. You've been extremely engaged. And in regards to sharing to your participant community, you'll see their names on a lot of the documents and resources. So congrats to both of you and thank you for being engaged and being here. Thank you. One more thing. Anybody else? Okay, we have a question in the queue here. I'm just going with the most votes here. The first one is a multi-question question. Part of it is what EHR system do you use now? And do you direct submit to ACC instead of using a third party software vendor? So I'm gonna take that question. Any of the six that I put in my presentation have the ability to do exactly what we do at Epic? Just so you know, CERN has the ability. Any of the six have the ability. It's directly uploaded from our EHR straight into ACC. But you're using Epic. We're using Epic. And I can do it daily. I could do it every five minutes truly. And I do sometimes. I'm a firm believer in what Lisa said. Do upload your data as often as possible. It is crucial to being able to look at your outcomes ahead of the quarter ending. So you can make adjudications. You can look at where you're going and where you want to go for the future. So yeah, it's a simple upload straight into ACC. And so our CAS PCI goes through. AFib, ICD, and that's the only, the first three that we've onboarded. We fully expect to do heart failure. That's the next one on our list. It's actually probably gonna be coming out very shortly. Heart failure, as we all know, is a huge, huge patient population. We all say it. Re-emission rates. And so having it in our EMR, a database in our EMR to really, again, use that AI component to pull information seamlessly, not using a third party, but something that's centralized will be, for now and in the future, instrumental in changing how we treat our patients. Truly. And we use Epic also. And I believe you said, Serena, it took about two years to integrate. The question is how long did it take to integrate? Two years, would you say? Two years. It was such a long process. Because it's not easy. It's changing a culture. Changing 24 hospitals who had their unique way of doing things, had seven different CVIS groups and seven different CVIS, you know, grandfathered in, not wanting to go to a one system. And so changing that thought process. Doing it two years, constant, every month, meeting with the key players, the data, the cath labs, the managers, the nursing units, truly listening and taking their input before we made that huge change. And so, yeah, two years. But I mean, I certainly think that we're not the only system that has done this. So believe it or not, there's four other major systems. And if you are, I applaud you because we stand behind you because you've been ahead of us. And so we've followed in your footsteps, those four larger institutions. But it was necessary for us all to be speaking the same language in our cath labs. One was McKesson, one was this, one was that. We now speak the same language and that same language goes into one data repository, so. Thank you for that. And then the next question, I think, speaks to also something you mentioned. And I think if you can just elaborate, right, that the project- I hope I can. The project you did was more of a pilot. So when we talk about the fields that are actually being pulled for AI or those discrete elements, you mentioned there were about 43, I believe. They're not a pilot. They were pilots six months ago, but now they are fully integrated. But the question- Go ahead. The question is more around what if there are discrepancies in the medical record? And the example used in this case is if in one area of the cath report it's diagnosed or it's noted as a STEMI, and then in another area of documentation it's noted as an NSTEMI. So it's more speaking to the types of fields that- So I'm gonna- I think that is whoever asked that excellent question. So those questions still need a higher thought process. They need a data abstractor. She mentioned hemoglobin, making sure we get the lowest hemoglobin. We've worked with our data abstractors. We know the definition. We worked with our IT to make sure that they were able to then make a algorithm to ensure that we got that lowest hemoglobin, that there was no discrepancy. And then we beta tested it for six months, flawlessly. So we know what the two together, we came up with a solution and we know what we're getting is the correct information. Now, going to a higher learning or a higher thought process, yes, that still is gonna require a data abstractor. It really is. And so we're not there. Where we are is in our infancy, 40 to 60 data elements, the most simplest of data elements that don't need an RN. And everyone on my team were advanced nurse practitioners. They don't need me to put a weight down. AI can do that. But they do need my expertise for what's a STEMI, what's not a STEMI, type two, type one, however it is. We're not there yet. Is the ability to get there? There is. I hate to tell everyone it's there. It is gonna be our future. But it really means changing the dynamics of how we do things currently. Our physicians, they're mirrored down in the old way of documenting, dictating, et cetera, even in EHR. And so it's changing that way of how they document. They don't need to make it a five page, you know, we can't, and so it's really changing that culture. And we're not there yet because of that. So little steps, we're doing little steps throughout Sutter Health. But when you take into consideration, we have so many registries. Even if I can lessen the burden by 10%, it is a cost savings that will equal at some point millions of dollars. And so it's definitely worth exploring. Thank you for that. That answered a couple questions. Okay. As you elaborated, I know talking about, you know, how you're validating and, you know, speaking to some of it requires that additional data abstractor to double check and. I was firmly against it. Can you imagine two years later and I'm on stage and I'm telling you it works? I wasn't. Two years ago, I said, how is this going to, because, and we caught a mistake in our AI, and that was for diabetes. I caught it a year ago. And I said, a patient who had never been diagnosed with diabetes ever came up in our AI system as being a diabetic because of one glucose level that was 125, but it was a non-fasting glucose. Again, it is a complete collaboration with an IT who knows how to program, but my brain power is going to teach him what he, how he's going to program it. And we now don't have that problems at all. Thank you. This is wonderful. This has been a wonderful discussion and I'm sorry that we are at the top of the hour already. So to be respectful of time, thank you very, very much. Thank you. And we appreciate you being here.

data abstraction

healthcare industry

AI approaches

artificial intelligence

machine learning algorithms

medical records

EHR-based data registry repository

manual approaches

abstractors

EMR system