Proactive care delivery: The AI-powered future of long-term care with Annette Salisbury, senior vice president of HCC clinical services at PruittHealth

Episode 79 January 24, 2024 00:26:43
Proactive care delivery: The AI-powered future of long-term care with Annette Salisbury, senior vice president of HCC clinical services at PruittHealth
The Post-Acute POV
Proactive care delivery: The AI-powered future of long-term care with Annette Salisbury, senior vice president of HCC clinical services at PruittHealth

Jan 24 2024 | 00:26:43


Show Notes


In this episode of the Post-Acute POV podcast, Annette Salisbury, senior vice president of HCC clinical services at PruittHealth, is joined by our own, Daniel Zhu, vice president of product management, data, analytics, and AI/ML, and Kelly Danielson, clinical product manager, to outline the future of artificial intelligence (AI) in long-term care and the current impact of proactive care delivery.

The trio shares their personal experiences using artificial intelligence and the benefits and challenges that come along with implementing these new technologies. They also address concerns that have prevented some healthcare providers from implementing AI in their organizations. Listen in to their discussion.

Topics discussed during today’s episode:

  1. [00:35 – 01:19]: The host introduces guests, Daniel Zhu, Kelly Danielson, and Annette Salisbury. Daniel details the key topics of the episode.
  2. [01:59 – 03:51]: Kelly and Annette discuss examples of artificial intelligence technologies and tools being used in long-term care.
  3. [04:34 – 06:03]: Annette addresses how she has helped implement AI in Pruitt’s facilities and how the technology has made a huge impact on patient care and family satisfaction.
  4. [06:11 – 09:02]: Annette outlines how beneficial artificial intelligence can be with the use of data, specifically in electronic health records.
  5. [09:55 – 12:33]: Annette shares how artificial intelligence is proactive and has helped her staff know which patients are at a higher fall risk.
  6. [13:12 – 16:27]: Annette explains the challenges she faced when first using artificial intelligence technology and how her staff reacted to the use of this innovation.
  7. [16:45 – 18:03]: Kelly and Annette discuss the fall program at PruittHealth.
  8. [ 18:29 – 20:46]: Daniel details how artificial intelligence will not replace humans but will drive efficiency and the efficacy of care.
  9. [21:28 – 23:47]: Daniel discusses how AI systems can be valuable for documentation processes and Annette outlines how AI systems are helpful for the nurses.
  10. [24:09 – 26:09]: Daniel summarizes the discussion and concludes the episode.



The content in this presentation or materials is for informational purposes only and is provided “as-is.” Information and views expressed herein, may change without notice. We encourage you to seek as appropriate, regulatory and legal advice on any of the matters covered in this presentation or materials.

©2023 by MatrixCare


View Full Transcript

Episode Transcript

Speaker 1 (00:02): Welcome to the Post-Acute Point of View podcast, our discussion hub for healthcare technology in the out-of-hospital space. Here, we talk about the latest news and views on trends and innovations that can impact the way post-acute care providers work. We'll also dive into how technology can make a difference in today's changing healthcare landscape for home and facility-based workers and the people they care for. Let's dive in. Speaker 2 (00:35): Now, without further ado, I'm pleased to introduce our speakers for today. First, we have Daniel Zhu, Vice President of Product Management, Data, Analytics, and AI/ML at MatrixCare. We also have Kelly Danielson, Clinical Product Manager at MatrixCare. And finally, we have Annette Salisbury, Senior Vice President of HCC Clinical Services at PruittHealth. Daniel Zhu (01:02): Today, we're going to go through, and Kelly and Annette are going to talk about when they implemented AI systems. And hopefully, we can address some of those concerns that's preventing some of the healthcare providers out there today from implementing AI at their facilities. So without further ado, Kelly, take it away. Kelly Danielson (01:19): Thanks, Daniel. Believe it or not, modern artificial intelligence is already used in a lot of the spaces in long-term care. Just off the top of my head, I can think of wound care. There's some advanced technologies, where you can take a picture of the wound, and it's able to measure the depth and the size, the area, of the wound, just from the image on your phone. There's also radar tech, that is able to detect and identify people in a room and determine if any of their actions can result in falls. So let's go to Annette. Annette, can you tell me a couple of good examples that you've seen or used in artificial intelligence in long-term care? Annette Salisbury (01:59): Yeah, Kelly, sure. There is AI technology out there currently in long-term care, and care coordination is optimal in patient coordination and making sure that that coordination of care is there, predicting risk and best practices out there in resident placements, especially identifying which facilities would provide the best care for patient conditions. We have monitoring systems for falls, rehospitalization, recognizing those activities, which room would be the best place to place a patient? When you're looking at an admission coming in, AI can help us do that, especially when we're looking at a hall with a patient population. AI is great for this, because now, I can look at patient coming in, and "Do I really want to put another patient that needs a lot of care on a hall that already has a heavy population there?" So this system is great for us, because looking at the acuity of those halls, I can say, "I really don't want to put another patient there, looking at my staffing." (03:03): So this is a great tool. Also, this is a predictor. It's real time. There's a lot of systems out there that we already currently have, but it's after the fact. Where AI tells us real time. It gives us the ability to predict what's coming, which that's what I love. I was the old school DHS, that used to have to dig through those paper charts. I had the paper across my desk trying to predict, "Is this patient going to fall? When are they going to fall?" Same thing with wounds. You mentioned wounds, Kelly, where we can now take that picture, look at that wound, send it to the doctor, and try to predict, "Is this wound going to get worse? Is it going to get better?" Before, we were trying to fly by the seat of our pants, where AI has that ability to help us predict these things and predict rehospitalization. So we do, we use this tool. Kelly Danielson (03:51): Awesome. Thank you. And it does empower the clinical staff and significantly impacts the clinical efficiencies in patient outcomes. And artificial intelligence powered clinical surveillance is definitely an extremely useful tool, as you've attested to. So nursing staff, nowadays, you have a dozen care providers entering clinical information into the EHR daily, and you're working to manage three or more dozen residents on your shift. And getting help keeping an eye on those residents would really be appreciated. And so, tell us, Annette, have you implemented any of those technologies like that? And if you did, how did it affect your organization? Annette Salisbury (04:34): Yes, Kelly, as I said, we've implemented AI in our facilities, and it absolutely has. It's made a huge difference. With the ability to implement this, it has given us a leg up, to help us predict falls, predict rehospitalizations. We use this in our morning clinical meetings. We use it in our weekly at-risk meetings. We've actually implemented it with our pharmacy consultants, so they actually are able to review drugs, to reduce our anti-psychotic medication. So this has helped us tremendously. Our physicians like it, because we have what we call our top 10 going into our weekends for rehospitalizations. Because we all know we don't want those rehospitalizations. And our physicians would have their top 10 list on paper, and then, we're like, "Oh, well, let's pull up AI and see what AI says." Because we would compare that list, and AI would have a completely almost different list than what the physicians would have on paper, that they would never think to look at. (05:30): So it's a great tool to predict and to look at those type of residents. So yes, we have implemented this, and it's been a great tool to help us manage those negative outcomes. It has also helped our family with family satisfactions as well, because it's been a realistic tool to help them understand why a patient is a fall risk, to help them identify those areas that they would not understand in the past of why that makes a patient a fall risk. So that's been a great tool as well, for family and satisfaction and outcomes as well. Kelly Danielson (06:03): That's amazing. Can you give us a little highlights, any key parts of the chart that artificial intelligence helps you with? Annette Salisbury (06:11): Yeah, as you said, Kelly, we input data in the electronic health record. And one thing nice about AI, and as I said, all of it's real time, so it collects that data as we enter stuff. So as your clinicians are entering, not just your nurses, but as your physicians are entering physician orders, as your nurses are entering data into assessments, as your CNAs or frontline staff, if you don't have CNAs, are entering your point of care, as they're entering those ADLs, as your nurses are completing your MDS', all that data is actively gathering on a real-time basis, so you're getting that real time. As I said, we've had systems in the past that it don't spit that data out until you submit that MDS, so it's after the fact. So you're still playing catch up. Where AI gives you that real time data. I love the fall risk, because as I preach to the staff, I could not be a fall risk at seven o'clock in the morning, but by two o'clock in the afternoon, that banner is there to tell my staff that I'm a fall risk. (07:16): So it constantly updating as the staff and my clinicians are putting input into the system. So it's constantly working and constantly updating as stuff is being inputted into the electronic health record. So as assessments are being completed, as point of care is being put into the system, as the physicians and the nurses are putting orders into the system, it's constantly working and generating reports. And at any point in time, as myself, I'm constantly in the system, so I monitor over 100 facilities, and I can go into any building I want to and I can look at the reporting and see what buildings have fall risk and look at the graph and see which halls have higher acuity. (08:03): I can have a facility call me about admission and say, "Hey, Annette, we're really wanting to admit." I'm like, "Well, which unit you're going to put that resident on? Because you really don't want to put them on your post-acute unit, even though I know that's where you'd like to admit, because look at the acuity there." When we do our meetings, we've incorporated this into our workflow, every workflow. I've had buildings call me and say, "How did you implement this?" Well, we just put it into our workflow. It is part of every piece of our workflow we incorporate AI into, to get it in there, because it is part of the workflow. You look at this, and it's just part of our workflow. Kelly Danielson (08:36): I like what you said about the difference between being in real time and MDS information. We like to say that MDS is like looking in the rear view mirror, but artificial intelligence puts that stuff there right in real time, so it makes a huge difference. Let's talk just a little bit more about the implementation process. Can you give us a little hint as to how that went or what you did to make it successful for your staff? Annette Salisbury (09:02): Yeah, Kelly, as I said, we put this in our workflow. When first implemented, everybody was like, "Oh, you're giving us another system. You're giving us something else to do." And we were like, "No, guys, this is part of your workflow." As I said, we implemented it into our morning clinical meeting, and we do a weekly at-risk meeting. I'm like, "Guys, this is just part of our process. You pull the charts, you're looking at your 24 hour reports, pull the system up, just pull it up, and look at it as part of your 24 hour review of your charts. You're going to see a change. When you're in your weekly meetings, just pull the system up. You have your system pulled up anyways, just pull this up. You can see it across the board." And once they started using it, they loved it. And then, as other departments heard, they're like, "Well, can we get access to this? Can we get access to that?" So it was like a chain reaction. (09:55): So my therapy department wanted access to it, because they're like, "We want to be proactive." So it was just like a chain reaction. So I trained our therapy department. So then, they start coming to morning meeting and saying, "Hey, we noticed so-and-so had some falls or they're high risk for falls or they're moderate risk for falls, can we go ahead and be proactive and pick them up?" So they're bringing a list now to morning meeting to say, "Can we pick this person up, this person, this..." And nursing's like, "But they've not had any falls." And they're like, "But they're moderate risk. Can we start picking them up?" So now, we're starting to see a chain reaction in our QM starting to decrease, because therapy's getting involved. And then, we have our own Medicare Advantage, and those physicians, through their medical record system, it case risk adjust their residents to, "This person is only seen monthly. This person is seen weekly." (10:47): Well, their people were hitting our radar report in AI differently than what it was hitting in their system. And they're like, "Wait a minute, can we get access to this report?" Because I have access to that report in our electronic health record. So I'm like, "Yeah." So I trained them as well. It was a chain reaction that everybody wanted access to this in our electronic health record. And even our RDs, because they're like, "Well, we can start picking up on people for weight loss before they even have weight loss." Because they started seeing the benefit that you could actually be proactive with what you're seeing. Because even though I don't have a wound yet or I don't have weight loss yet, if I'm starting to see a decline in my mobility and I'm starting to see maybe an increase in pain, well, we know that, if I have pain, I'm not moving, then what's going to come next? (11:37): I'm going to start to break down. I'm going to have maybe some depression going on or, if I've already got some depression, I'm not getting out of bed, so this is what's going to start to happen. So I can start being proactive on my end and start putting interventions in place to prevent weight loss, to prevent the pressure ulcer. So it's a very proactive tool, and it's real time. It's going to help you prevent your QMs from skyrocketing. It's going to do the right thing for the patient, because it's the right thing to do. (12:06): You're going to get all your interventions in place, and then, for the fall risk, you're going to catch them before they hopefully fall. You're not going to prevent fall, because if the patient's going to fall, they're going to fall, they're going to get out of that bed. One way or the other, they're going to fall. But at least you're going to know that they're fall risk. You can hopefully put some interventions in place and you can educate your staff and your family members of why this person is at high risk for fall, because it helps you identify that. Kelly Danielson (12:33): That makes sense. You made a comment earlier about that, "Oh, you're adding more work to us," when you first implemented it, and that's what they thought. But intelligence systems today are being built right into what we're used to doing, and it does everything. Daniel referred to autocorrect. Remember when it first came out, it doesn't impact anyone who doesn't use it, but if you've gotten used to using it, it fits so well with how we text and search now. So tell me about your teams. When you first started with your artificial intelligence technology, what were the challenges of getting the buy-in or getting them to see a benefit from it? Annette Salisbury (13:12): I think, once they realized that it wasn't an added addition to what they were already doing and they understood the why, I learned this a long time ago that you just can't throw something at somebody unless they understood the why. And I think, once they understood the why and it wasn't adding additional work to their already load, and we just added it into their already workflow, that they were fine, they were fine, that this was actually going to help them, and it was preventative, that heck, if I could prevent that fall and get something in place and that would be preventing me from having to do an incident report and do additional work, that it was great. They loved it. And my directors of nursing were like, "Oh my gosh, this is going to prevent me from sitting at my desk and trying to do the old paper way of having all the papers spread out and trying to track and trend and figure out where that next fall was coming from." They bought into it very quickly. Kelly Danielson (14:14): That's amazing. I was a corporate nurse for a long time too, and overseeing buildings, and as a nursing leader, to be able to push that out to an entire corporation is pretty significant, Annette, so it's also very impressive. So as you pushed it out to all of your corporation, was there one thing that was just like crazy? Or what did you have to do to make sure that everybody got the memo, if you know what I mean? Annette Salisbury (14:45): I think it's just the redundancy, Kelly, and the teaching. And all my nurse consultants use it and getting them to continue to use it. So every teaching that we do, every education that we do, AI is part of it. So we all speak the same language, and as I said, we have 100 plus [inaudible 00:15:04] facilities, so we all try to teach the same thing. We try to keep the consistency. So when we speak rehospitalization, and I have a rehospitalization nurse, the first thing she said is, "Did you pull the AI up? Did you pull this up? Did you look at this before you sent this patient out? Did they trigger?" When we roll our falls program out, it's part of our falls program. "So did you see, did they trigger high risk? Did you go back? Did you look? Well, why?" Now, I got my pharmacy consultants speaking the same thing. [NEW_PARAGRAPH]"Did you look to see, did they trigger fall risk? Did you look at their changes in medication? Did they trigger on there for changes in anti-psychotics?" Because that part of the AI, when you look across for their falls, it looked at the anti-psychotics. "How many anti-psychotic medications have they had? Were they changed in the last 30 days? What was the last dose?" And it even includes their PRNs there. So we were all preaching off the same thought [inaudible 00:16:01]. So I think that consistency, with even us corporate staff, when we go into the facilities, we're all singing off the same page. With every training that we do, it's written in our policies and procedures to a certain degree, or at least our procedures, and our workflows. It's there. So we don't deviate, and they're hearing the same message over and over and over and over again. And that helps. Kelly Danielson (16:27): Yeah, consistency for sure. That's awesome. I just want to delve a little bit more into the falls risk for you guys. You and I have talked before about how it's gone for you, but can you just tell me a little bit more about the falls program and any successes that you have there? It's so interesting. Annette Salisbury (16:45): Yeah. So falls has been a big project for us. Falls has been on our radar for over a year, and we just redeployed our falls program and falls implementation. And last month was Falls Initiative Month for us as well. And last February, if you compared last February to the February before, a year ago, and just looking at those numbers, and that month alone, if you compared those numbers to the numbers a year ago, we decreased falls by almost a thousand, in just that one month, just a month to a month. So a month for February to February a year ago. (17:22): Now, every month is different. We go up, we go down, we go up, we go down. But that month was significant. I don't know what happened that month, if it was a fluke or what, but that was very significant for us. And we are getting ready to redeploy our... As we said, last month was, it was actually a National Falls Month, so we had a big campaign, we preached it. The whole month was falls, falls, falls, and AI is a big part of that. So we talked about that in our core value every day. "Have you been reviewing your radar, the fall risk?" All that kind of stuff. So just the constant education. Kelly Danielson (18:03): That's awesome. We know that every caregiver in this industry, for any significant length of time, is in it to take excellent care of our seniors. We know that. As nurses, that was our goal. That's why we went into nurses training. So this is a great way to highlight the intelligent technologies. We know that they're not there to replace anybody, but rather to drive efficiency and effectiveness of care. That's already going on in long-term care. In fact, I think we're only going to see more. Don't you agree, Annette? Annette Salisbury (18:29): I do agree with that, Kelly. Kelly Danielson (18:31): Yep. Okay, Daniel? Daniel Zhu (18:34): I completely agree. I think, as we go into the future of long-term care and intelligent systems, it's only going to get more and more integrated with the workflows, and it's going to be more ubiquitous in the overall healthcare space. We're just going to see more and more devices, I believe, help drive that efficiency, like you said, Kelly, help drive that effectiveness of each individual healthcare provider and caregiver out there, just to make their jobs easier, so that they could help more individuals at the same time. And AI is just going to get more and more powerful, as you add more devices into the workplace, because these devices, just like that caregiver right now, who's documenting in the EHR, entering the vitals from time to time, whenever they do it, entering documentation, these devices are going to be monitoring your residents, monitoring your patients constantly. (19:25): And all that information generated is only going to be more and more fuel for all these intelligent systems out there, and it's going to drive towards a more effective AI system. So its predictions are only going to get more accurate, and at the same time, it's going to make the job easier for all the clinical staff there, just so that it can keep an eye out on all these individuals at your facilities at all times, something that Annette alluded to as well, about how it's always working. Even if you step away, even if some of your caregivers aren't there, it's always working, always keeping an eye out for the residents and making sure that their outcomes are cared for. Annette, on your side, what do you think is going to be the future of these intelligent systems in long-term care? Annette Salisbury (20:12): Daniel, I completely agree with you. I just see AI in the healthcare system just taking off. I 100% agree with you. I see it helping us in the healthcare system. I see it just skyrocketing and decreasing the workload. I'm a nurse, just like Kelly is. I'm excited for this. Decreasing the workload, just being there to help us with data, documentation. Documentation is a burden. That's one thing that we stress over. Sometimes, we do the care, we provide good care, but our documentation is lack. (20:46): I see AI helping us with that tremendously in the future, gathering that data, gathering our vital signs, helping our CNAs, our frontline staff. That's another thing that we struggle with. Those CNAs are out there doing that care, providing that care, but we lack in gathering that documentation. I see AI helping us tremendously with that aspect as well, with gathering that information, even further with that, maybe with the voice activated, with gathering that information, helping us gathering the point of care documentation even further with AI. So yes, I think AI is just, way in the future, not even way in the future, here today, going to help us tremendously. Daniel Zhu (21:28): Yeah, I couldn't agree more. I think you brought up a really good point about documentation. Many different forms of documentation, just being able to record the session, the interactions between the caregivers and the residents, and making sure that we take the proper notes, so that there's proper follow-up, if required. Like you said, vitals and all those properly documented by AI systems, as well as, like you said, some of the nursing assistants, making sure that they're properly doing their roles and, at the same time, making sure that it's all documented properly. So I've seen some systems in the very preliminary phases, where it does video documentation and it can identify, "This person did this, this, and this," and right before they walk out the door, the system can even remind them, "Oh, you might've missed this task." If they actually missed the task, they can quickly go ahead, return, finish the task before they leave the room. (22:17): If they didn't miss the task, all they have to say is, "Nope, I already did it." And then, that just adds to the documentation system automatically. So that kind of point of care access to the residents and point of care access to documentation really allows you to intervene much earlier. So it takes that proactive, even the step earlier and say, "Hey, you might've missed something," or, "Hey, have you considered doing this with your resident?" Just because they have these other things that you might've missed during chart reviews or before you approach the residents. All these kind of spot on capabilities will only increase, like I said, as we add devices more and devices and more technology into the healthcare space. Are there any particular types of systems you're watching out for or you feel like will be personally beneficial to you and your team as you go forward, Annette? Annette Salisbury (23:06): Yeah, just the increase in the technology and, as you said, care planning, interventions, things like that. It helps the nurses. That's something we're looking for in the future. As you said, the interventions, ideas for the nurses moving forward, to make sure that they know, are given the ideas, because a lot of times, they're critical thinking. And I think that's where AI would help them, gathering that information and saying, "Hey, did you think about this?" Because a lot of times, the nurses are so busy doing the patient care, and the ratios, because we're talking about staffing, with the nursing shortage, sometimes, you have a ratio of one to 30, one to 44, and they're busy taking care of those residents and getting done what they need to get done. (23:47): And when they sit down to do that charting, they're just brain dead. So the AI, I think, would help them tremendously when they're doing that charting and help them remember, as you said, "Did you think about this? Did you think about that?" Or when they're inputting that information, it's gathering that information on a real time, that it would help spark different things like that. Daniel Zhu (24:09): Absolutely. I'm going to summarize today's session. AI systems are a variety of technology tools that utilize data to generate patterns and help facilitate with a specific task. That was kind of the overall first section of today's conversation of what AI tools are and how they work. And overall, they are particularly effective when there's plenty of data, plenty of information available, so that they can effectively kind of work and come to these solutions, come to these monitoring systems, to be able to sift through all of this information. AI is particularly good at that. And overall, AI is designed to compliment the work that your organization and your caregivers do. And it's not really there to feel like additional chore, as you implement one of these systems, but rather just drive efficiency, in the way that the caregivers are currently doing day-to-day work. Overall, it should feel natural, to be alongside your day-to-day work and the tools that you use today. (25:09): On a high level, these systems still require some adoption and implementation considerations. I think Annette has brought up some really good points, that there are some challenges. You can't just throw technology against the wall and hope it'll stick, but making sure your team is aware of the overall benefits and making sure there's a consistent messaging with your teams on the long run. And it really helps you achieve the goals that you set out to achieve by implementing these systems. So that's really important to have a goal when you're implementing these systems, not just for the sake of implementing a system, but rather, like Annette said, either you're looking for falls, you're looking for psychotropic meds, have a goal in mind, and then, reiterate with an AI system. And then, finally, AI systems and tools and advanced technology is only going to increase in the long-term care space. So of course, taking the first step in learning how to integrate and better work with these tools today can really improve your overall capabilities as caregivers. Any last thoughts or anything I might've missed, Annette? Annette Salisbury (26:09): No, I think you hit it right on the head. Daniel Zhu (26:12): Awesome. Speaker 1 (26:15): That concludes the latest episode of the Post-Acute Point of View podcast. We have a lot of guests and topics coming up, that you won't want to miss, so be sure to subscribe. To learn more about MatrixCare and our solutions and services, visit You can also follow us on LinkedIn, Twitter, and Facebook. Thank you for listening. Be well, and we'll see you next time.

Other Episodes

Episode 1

August 27, 2018 00:24:25
Episode Cover

Disrupting the Status Quo

In this episode of MatrixCare's podcast, we introduce Navin Gupta – VP of Software Engineer.  Gupta analyzes the trends, technological advances and disruptive innovations...


Episode 8

June 12, 2019 00:43:30
Episode Cover

How to impact culture in the world of technology

In this episode of MatrixCare’s podcast, we introduce Navin Gupta – VP of Home Care Solutions for MatrixCare and Jerry Bridge – Professional Motivational...


Episode 85

May 08, 2024 00:44:07
Episode Cover

Care coordination playbook: Optimizing referral relationships with Deb Wesley, CEO of Addison County Home Health and Hospice

In this recent episode of the Post-Acute POV podcast, our host, Nick Knowlton, VP of strategic initiatives, has a conversation with experts Deb Wesley,...