00;00;00;00 - 00;00;24;13
Bryan Nowicki
Hello and welcome to Hospice Insights: The Law and Beyond, where we connect you to what matters in the ever-changing world of hospice and palliative care. Getting Ahead of the Curve: How Hospice Providers Can Build Smart AI Practices. Taylor, how are you doing? Welcome back to the podcast.
00;00;24;16 - 00;00;29;29
Taylor Crossley
Thank you so much for having me a second time. This has been a lot of fun. So thanks for thanks for inviting me again.
00;00;30;01 - 00;00;52;15
Bryan Nowicki
Oh, sure. And not only is this a second time, this is a two-parter. We're going to be talking about this, really it's a it's a topic that is taken over almost every industry, business, health care provider over the past two years. And it kind of hit me like this happened all of a sudden, like out of nowhere, everybody was talking
00;00;52;15 - 00;01;27;29
Bryan Nowicki
AI and it went from an afterthought to front and center, and everybody is trying to pursue it. We know the government is making use of AI, you know, in a way that's not necessarily making life easier for hospices and home health agencies. Right. But but everybody is using it. And so this is very timely because, Taylor, what what you have done is you've kind of, developed a bit of an expertise in AI in the health care space.
00;01;28;01 - 00;01;51;18
Bryan Nowicki
And in this two-part series, you're going to share information about, you know, what what to, how to manage the use of AI, how to implement it, where it's being used, ways that you can use it effectively. And we'll also talk along the way about some areas where you might be pushing the envelope a bit, and maybe you should be cautious in the use of AI.
00;01;51;21 - 00;02;14;25
Bryan Nowicki
So in this part one, we're going to lay the the foundation for all of this, in part two then which, which we’ll we’ll, post up a few weeks later. We're going to get down to some specific practical applications. So. First of all, Taylor, what what caused you to have an interest in AI and kind of jump into this with both feet?
00;02;14;26 - 00;02;35;26
Taylor Crossley
Yeah, I think that AI is a really interesting area where providers can like, you know, I introduced myself last time. One thing I really love doing is helping providers, you know, innovate. And this is one area where providers can really innovate their patient care, by using AI. But there also needs to be the safety mechanisms around it.
00;02;35;26 - 00;03;03;17
Taylor Crossley
So I really jumped in because this is an area of privacy, that I have a specific interest in. And then also this overlap of innovation, providers can do really cool things with AI in hospice. And so I think that that's really why I was so interested in learning more about it. It's such a new and evolving area and something providers really need to be paying attention to because like you said, I mean, it's been probably six, eight months
00;03;03;20 - 00;03;12;18
Taylor Crossley
when this first started to percolate and it's percolated very fast. So it's something that I wanted to keep my pulse on and something I find really interesting.
00;03;12;19 - 00;03;27;03
Bryan Nowicki
Great, great. Well, let's jump into this. You know, when we talk about AI in hospice, are we talking about something that's that's emerging or is it already here? Is there still evolution going on. What's the current status.
00;03;27;06 - 00;03;52;01
Taylor Crossley
Yeah, yeah. Like I kind of previewed it's already here. Whether the hospice organization really knows it or not. In fact, many hospice organizations are using AI features that they maybe never even specifically adopted, but are just embedded in other vendor programs that they have already and providing their patient care. So, you know, the data makes this even more clear.
00;03;52;07 - 00;04;25;05
Taylor Crossley
67% of health care executives report investments in AI to enhance patient care. But only 13% have a clear strategy for integrating AI into clinical workflows. So we're kind of in this, AI is here. It's it's not emerging anymore. It's something that's already arrived. But the governance around it is really what's kind of fallen behind. And so that's something that we should, you know, really be paying attention to here at the beginning of adopting AI into our organization.
00;04;25;07 - 00;04;44;21
Taylor Crossley
And in fact, 49% of health care executives, name appropriate use of AI as one of their top three challenges right now. So that's the real gap that we're dealing with right now. AI is not a future problem to plan for. It's a present day compliance and liability issue to manage. Yeah.
00;04;44;21 - 00;05;11;19
Bryan Nowicki
And I think AI is always improving. I mean it you can tell in just how it learns from how I use AI. I'm sure how you use AI, it learns and it gets better. And all these AI features are going to do the same. But if you're not on top of it now and you try to get in a later, you know, you might be missing a lot of the learning process about how to use AI effectively and what are some of the pitfalls.
00;05;11;20 - 00;05;37;03
Bryan Nowicki
So yeah, you're not seriously exploring how I can help your organization today. You're already behind and you don't want to play catch up. Going into the future, you want you want to make sure you grasp on to these opportunities and understand some of the concerns. So, yeah, let's get specific then. Tailor, you know, where is AI in use in hospices today?
00;05;37;10 - 00;06;01;10
Taylor Crossley
Yeah. One area is clinical documentation. That's the most prominent use case that I'm seeing right now. Providers using ambient scribes. Which are tools that listen to a patient encounter and generate a draft clinical note. That's something that I'm seeing a lot of providers using and something that can be built directly into EHR platforms.
00;06;01;12 - 00;06;30;23
Taylor Crossley
So maybe it's not even that the hospice organization themselves has, you know, maybe bought an ambient listening tool, and incorporated it into their, their process and procedures, but rather maybe it's that EHR platform that you've been using forever has integrated AI, into its platform. So hospice nurse nurses and other providers may be using them without the organization having any policy in place.
00;06;30;26 - 00;07;07;23
Bryan Nowicki
And this is an area I'm glad you raised this one first, because I think I, and maybe it's my attorney paranoia, but I, approach this rather cautiously because there are times when kind of the use of AI may bump up against expectations laid down in regulations, that the most obvious example is for hospice is the, physician narrative that's included in a certification of terminal illness, where the physician has to attest that he or she compose the narrative.
00;07;07;26 - 00;07;37;19
Bryan Nowicki
If you have AI doing a draft narrative, does the physician really compose it? And I think there's arguments both ways physician can lay eyes on it. Edited, revise it, know how much editing and revision does it take for that to be the heart of the physician's own composition? Yeah, I think that still has yet to be worked out from a regulatory side, and it's not clear how far CMS may allow that to happen.
00;07;37;21 - 00;07;55;24
Bryan Nowicki
But I think that's a word of caution. AI can be very helpful in making documentation more efficient. But what are the lines that exist in the regulations that might put some limitations on that? Anyway, that's that's what give me the heartburn on this particular issue anyway.
00;07;55;26 - 00;08;24;26
Taylor Crossley
Yeah. Yeah. I mean that's that's a really good point that where are those clear lines and the regulations don't set them. And I don't think that there's been enough use like like we were talking about this is like, six months last year, you know, that, providers are really starting to adopt, these kind of tools. And so it's really questionable about when, you know, where is that line, for the physician narrative, I wonder if a solution is to use the AI listening tool during the face to face encounter.
00;08;24;28 - 00;08;48;13
Taylor Crossley
You know, from that third benefit period on, be able to use the listening tool for those clinical notes. But I think at this point, it's probably best practice from a risk management perspective to take those notes and make sure that your physician is still composing it themselves, because I think that is going to be a, a big question of what is the definition of compose.
00;08;48;15 - 00;08;58;19
Taylor Crossley
And when you've got your physician attesting to having composed the narrative, I would advise staying on the, the safer side, being a little bit more risk averse.
00;08;58;25 - 00;09;19;14
Bryan Nowicki
And, Taylor, I think attorneys are often quite risk averse or at least like to point out the most conservative option. And of course, our clients make the final call. But yeah, I that that's, we're good at spotting issues and risks, and this is certainly one of them. But what other areas do we see? I being used today in hospices.
00;09;19;15 - 00;09;52;11
Taylor Crossley
Yeah. Another another area is coding and billing. So hospice billing software often or increasingly is embedding coding assistance and documentation alerts. These are often frames as compliance or quality features. And may not be labeled prominently as AI, but may use AI to help those functionalities do the work that it's doing. So that's kind of the hidden AI issue and something to be paying attention to when you're and you're coding and billing software and.
00;09;52;15 - 00;09;53;24
Bryan Nowicki
Thought about it. Yeah. Go ahead.
00;09;53;25 - 00;09;55;06
Taylor Crossley
No go ahead Bryan. Sorry.
00;09;55;10 - 00;10;10;04
Bryan Nowicki
Well and what about the compliance side is, is a, a tool that, you're not not providing real time documentation assistance. But what about the compliance folks out there? Are they using AI?
00;10;10;05 - 00;10;36;10
Taylor Crossley
Yeah, we're starting to see compliance folks use AI. Whenever they're going through documentation, maybe in an audit. Or maybe they're just not even in an audit yet. But they're about to submit records for a claim for a patient. And compliance can use these tools to maybe run their records through run a physician narrative through an AI software tool to see.
00;10;36;13 - 00;11;11;02
Taylor Crossley
Are these physician narratives sufficient? Do they meet the requirements under the regulations? Do our plans of care have everything that they require per the regulations? So this might be an area where compliance can use AI as a pre billing review or a pre billing audit. To kind of preview where there are maybe some places for improvement or maybe where they can educate their providers, and their staff on the requirements for documentation, for submitting claims and other area.
00;11;11;04 - 00;11;40;08
Taylor Crossley
That we're really seeing is, family communication. So once we get through that kind of, you know, clinical use, you know, we're using it for billing. There's also the patient side of using AI. So maybe you're using AI to communicate with family. Maybe you're using an AI assisted scheduling tool. Maybe you're using an AI assisted update tool or chat box, that are showing up in a patient and family facing interaction.
00;11;40;10 - 00;12;02;10
Taylor Crossley
So in the hospice context where family communication is core to the care model, this is one area that's certainly worth watching closely as well. The kind of final area that I'd really highlight for providers. You know, you've got you're using it in clinical documentation. You're using it for billing. You're using it to communicate with patients in a patient facing manner.
00;12;02;13 - 00;12;36;09
Taylor Crossley
But then there's also this shadow AI use. I kind of like to call it, maybe you've got staff members who are using consumer tools like ChatGPT to draft notes, summarize patient records, or write communications to family members. And oftentimes because this is moving so quickly, right. They might be doing this without any organizational policy and often without, you know, an understanding of the HIPAA implications of inputting this patient data into a software like ChatGPT.
00;12;36;16 - 00;12;49;16
Taylor Crossley
And so that's really where organizations can kind of catch up. And the focus I think that we're going to be talking about today is the governance around AI. Using your staff is a really important place to be thinking about as well.
00;12;49;18 - 00;13;14;03
Bryan Nowicki
Well, that's concerning. Although, you know, there's certainly a lot of temptation with AI, it is so accessible. And although it might be, intimidate into old folks like me at first, once you get into it, it's can be addictive and become a crutch. So I can certainly see how this shadow AI issue, as you've described it, is, is a real problem or it could be a real problem.
00;13;14;05 - 00;13;15;19
Taylor Crossley
So yeah.
00;13;15;21 - 00;13;26;14
Bryan Nowicki
What do what? Let's dig a bit deeper into that. What are you seeing with shadow AI use and how the government or hospices have responded?
00;13;26;17 - 00;13;45;29
Taylor Crossley
Yeah. Yeah. Like you like you mentioned, this is a real topic and has been a big topic of discussion in the industry over the past year that AI is really starting to get integrated. So, you know, if you take, for example, a hospice nurse who uses ChatGPT to draft a visit note using a patient's name and clinical details.
00;13;46;01 - 00;14;09;13
Taylor Crossley
That's a situation where she has been sent to a third party with potentially no business associate agreement, and that's potentially a HIPAA breach. So that's the kind of example that we want to be thinking about. When we're talking about this shadow AI problem in hospice. And in fact we've seen recent OCR guidance. So HHS through the Office of Civil Rights.
00;14;09;13 - 00;14;33;28
Taylor Crossley
So HHS, OCR is the agency that helps enforce HIPAA. And they recently produced some guidance that made it clear that organizations are responsible for AI use across their enterprise, whether it's approved or not approved, which is prompting many providers to take a hard look at where AI is already being used and how to govern it.
00;14;34;01 - 00;14;58;04
Taylor Crossley
And so what we really suggest is prioritizing AI transparency within your staff. Most employees who are using AI, it's not malicious, right. Like this shadow use. It's not like there's bad intention by the staff members to be using AI. They're just really trying to work more efficiently. That said, right intent doesn't change the legal exposure.
00;14;58;10 - 00;15;20;01
Taylor Crossley
So that's why it's really important for hospices and other providers to have an acceptable use policy that explicitly addresses consumer AI tools. And having that in place to make sure that you're kind of addressing that shadow, I use policy and bring it out into the light. No more shadow use. You know what's going on in your organization?
00;15;20;04 - 00;15;46;02
Bryan Nowicki
Yes, exactly, exactly. And you use the term governance, or you kind of being able to monitor, have some control or directive out there. So when it when it comes to kind of building this government governance structure, what should the leadership of a provider and the folks who would construct that framework, what should be they be thinking about what needs to be put in place?
00;15;46;02 - 00;16;10;29
Taylor Crossley
Yeah. So starting with the basics, right. There's three main questions that you should be asking yourself as a leader of a hospice organization. First is what is your organization doing around AI? How is it using AI in the organization? You know, make those determinations of, you know, where is it being used? Just make a list is really my recommendation.
00;16;11;04 - 00;16;43;10
Taylor Crossley
Knowing where AI is being deployed in your organization is your strongest, risk mitigation tool. That second kind of question is asking whether your board or your executive leadership has enough AI literacy to exercise meaningful oversight over the uses of AI that you've identified within your organization. And then the final kind of big question to be asking yourself is whether your board has a structured AI education program or is everyone kind of learning on their own?
00;16;43;13 - 00;17;09;23
Taylor Crossley
The answers to those questions can guide the organization's next steps. So first, you've got to know where AI is deployed in your organization. You can't govern what you haven't inventoried, right? So start with that simple question what AI tools are in use? Who approved them, and which ones touch patient data? You know, include in your inventory, vendor software, embedded features.
00;17;09;29 - 00;17;30;22
Taylor Crossley
And then also that shadow AI informal staff use. And then second, once you have the answer to that question, you can know what is happening to the hospice's data. You know, which vendors have access to protected health information. You should know what they're doing with it. And you should make sure that your organization's contracts are clear.
00;17;30;22 - 00;17;56;21
Taylor Crossley
That patient data cannot be used to train AI models. That third kind of step is writing an acceptable use policy. So once you've identified that shadow use problem, you know you've got your employees are using ChatGPT to, write messages to family member as well. Then you can set kind of clear rules about what staff can and cannot do with AI tools.
00;17;56;24 - 00;18;19;25
Taylor Crossley
Especially consumer tools. Once you have inventoried how they're using those tools, maybe your organization makes ChatGPT off limits for anything involving patient data. That needs to be explicit in writing, and there needs to be training to back it up with your staff. The fourth kind of thing that we recommend is building a vendor vetting checklist.
00;18;19;25 - 00;18;47;16
Taylor Crossley
And use it for every new contract. So there should be AI specific questions, as a standard part of your, negotiating contracts process. So whenever you are assessing a new vendor, and even when you're doing your annual security risk assessment, that's a good time to kind of take an inventory of your vendors. Figure out the kind of AI that they're using in their platform.
00;18;47;18 - 00;19;07;15
Taylor Crossley
And make sure that it's up to snuff under the HIPAA security rule and privacy rule. But then again, using that vendor vetting checklist, not only, you know, now assess what we have, but in the future, for every new vendor contract that you might enter into. And we can talk more about that, I think, in part two.
00;19;07;17 - 00;19;33;21
Taylor Crossley
And then another thing to think about is ensuring human oversight over every clinical decision I touches. So I can assist, but it cannot be responsible for clinical decision making. Every output of AI, that affects patient care or billing needs to have a human who reviews it and be accountable for that. Use. And that kind of goes back to that acceptable use policy as well.
00;19;33;22 - 00;19;57;13
Taylor Crossley
And then the last piece is the most important. In my opinion, is embedding AI risk into the organization's existing risk management program. So this doesn't require a new department. We can treat AI as, you know, really the same as any other enterprise risk. AI isn't something new. This is something I've talked to with other people about as well.
00;19;57;13 - 00;20;27;09
Taylor Crossley
Just AI is just another technology, right? 20 years ago, we were all up in arms about the electronic health record, and providers were scared about what that meant. And so EHR was, a technological, advancement that scared a lot of providers, scared the industry. AI is pretty much exactly the same. You know, if you just compare those two things, you need to be able to assess the risks, identify it, and then manage those risks.
00;20;27;10 - 00;20;44;27
Taylor Crossley
So having an incident response plan that explicitly covers AI scenario, is a part of basic governance now. It's not really an advanced app. So making sure that you've got your IRM, your enterprise risk management heat map and AI is a part of that, assessment.
00;20;44;28 - 00;21;11;22
Bryan Nowicki
Great. Well, thanks, Taylor. And I know you know it. Every, every industry or community needs to address all these technical, technological advances. And I think AI's the most interesting yet. I remember, as an attorney trying to get used to using a dictaphone as a young attorney. That was a big deal. And then computers and then smartphones and now AI, which kind of blows it all away.
00;21;11;22 - 00;21;36;07
Bryan Nowicki
And I'm sure the same is true in the health care field as well, going from paper records to EMR. And now I, supposed to make our lives easier, but it only happens if you use it the right way. Have the right plan around it. Taylor, thank you so much for kind of defining the foundation. And, yeah, I'm looking forward to part two where we'll get into some specifics.
00;21;36;08 - 00;21;38;12
Bryan Nowicki
Vendor contracting among them.
00;21;38;14 - 00;21;39;06
Taylor Crossley
Yeah.
00;21;39;08 - 00;21;43;24
Bryan Nowicki
And yeah, we'll we'll talk again soon. So thank you very much, Taylor I appreciate it.
00;21;44;01 - 00;21;57;06
Taylor Crossley
Yeah, absolutely. I'm happy to start previewing these things that organizations can be thinking about. And I'm excited to, you know, dig deeper into each of these specific issues that they can and, you know, be considering moving forward. So looking forward to the next next time.
00;21;57;06 - 00;22;20;08
Bryan Nowicki
All right. Take care. That's it for today's episode of Hospice Insights: The Law and Beyond. Thank you for joining the conversation. To subscribe to our podcast, visit our website at huschblackwell.com or sign up wherever you get your podcasts. Until next time, take care.