WEBVTT
00:00:04.719 --> 00:00:10.080
This episode of AHLA Speaking of Health Law is brought to you by AHLA members and donors like you.
00:00:10.320 --> 00:00:13.679
For more information, visit AmericanHealthlaw.org.
00:00:17.199 --> 00:00:22.559
So, what are you really buying when you contract for AI-enabled digital health tools?
00:00:22.719 --> 00:00:42.640
This is the question that we're going to be discussing today, and especially in the context of things like AI moving into a more regulated space, the growth of remote patient monitoring and the potential changes to reimbursement in the regulatory atmosphere for RPM, as well as interoperability and patient engagement challenges as well as strategies.
00:00:42.880 --> 00:00:48.320
So one of the biggest things and challenges in the in the industry right now is fragmentation.
00:00:48.399 --> 00:00:58.159
And that's fragmentation in a couple of different ways, EHRs and different data sources and getting those things to talk to talk to each other, as well as fragmentation, I guess, among different states.
00:00:58.320 --> 00:01:03.920
There's so many different state laws that regulate this differently, as well as the federal government kind of as an overlay on top.
00:01:04.079 --> 00:01:26.879
So our job really as lawyers that are advising these companies, that are maybe investing in healthcare technology or partnering with these technology companies, is to turn this regulatory risk, turn these operational risks into diligence questions, into contract terms, into policies and procedures and training for people that are using these tools to ensure it's being done safely and responsibly.
00:01:27.040 --> 00:01:29.040
So that's essentially what we're going to be talking about today.
00:01:29.120 --> 00:01:35.359
I'm Sean Sullivan, and I'm a healthcare regulatory attorney and a partner at All Sinner Bird in Atlanta.
00:01:35.519 --> 00:01:41.120
I am joined today by Elisa Chessler and Betsy Hodge, and welcome to this AHLA podcast.
00:01:41.439 --> 00:01:50.480
I will note that this podcast is based on our session from HLA's 2026 Transactions Conference in Nashville from back in April.
00:01:50.719 --> 00:01:53.040
But let me turn it over to Elisa and Betsy.
00:01:53.120 --> 00:02:04.799
I'd like you each to each to introduce yourselves and I guess you know answer or talk briefly about the question: what are you really buying when you contract for an AI-enabled or some other advanced digital health tool?
00:02:05.200 --> 00:02:06.480
Elisa, do you want to go first?
00:02:07.040 --> 00:02:08.000
Thanks, Sean.
00:02:08.400 --> 00:02:09.599
Happy to be with everyone today.
00:02:09.680 --> 00:02:11.840
My name is Elisa Chester, as mentioned before.
00:02:12.000 --> 00:02:17.599
I'm a shareholder at Baker Donaldson based in Nashville and Washington, DC.
00:02:18.240 --> 00:02:55.360
And working a lot with clients on the adoption of AI tools and thinking about the best way to make sure that we are holistically thinking through not just those short-term issues of how is it working effectively now, but thinking through where are we going to find ourselves in the integration of those tools and where are we going to find ourselves in five years and 10 years with all the information that the provider community is going to be ingesting and uh dealing with.
00:02:58.639 --> 00:03:00.240
And I'm Betsy Hodge.
00:03:00.319 --> 00:03:05.599
I'm a partner at Akerman LLP in our West Palm Beach office.
00:03:05.759 --> 00:03:26.719
I'm in our healthcare practice group, and I focus on um the provider side of um purchasing um and investigating um these RPM tools and how they can improve my clients' uh care and also help their bottom line.
00:03:26.879 --> 00:03:29.439
So what about the what about the question?
00:03:29.520 --> 00:03:44.319
You know, what do what do what should clients be thinking about, or what should our clients, or if you're in-house, what are you thinking about when you're contracting for these AI-enabled tools or these, you know, other advanced digital health tools like remote patient monitoring platforms or interoperable uh software.
00:03:44.479 --> 00:03:45.599
What are you guys thinking?
00:03:46.000 --> 00:03:48.319
You know, it's a it's a software tool.
00:03:48.479 --> 00:04:00.400
And so a lot of times the contracting folks are looking at this as another software um application, but it's you know, RPM is more than that.
00:04:00.560 --> 00:04:08.000
Um, you know, as I mentioned before, you're also looking for operation something that can deliver operational efficiency.
00:04:08.240 --> 00:04:12.400
Um, and um Sean, you and Lisa touched on this too.
00:04:12.560 --> 00:04:32.399
Also um getting clinical insights and decision support from all the data that is being collected, and which then creates some issues around how do you safeguard that data and you know um responsibly use that data.
00:04:32.560 --> 00:04:36.720
Um, you know, and you also have some reimbursement opportunities.
00:04:36.800 --> 00:04:40.879
We'll touch on that, I think, a little bit later in the conversation.
00:04:41.360 --> 00:04:44.560
Um and risk transfer.
00:04:45.120 --> 00:04:57.680
Um I think so I think those are um some of the things that you're buying when you're buying um when you contract for you know AI enabled digital health.
00:04:57.759 --> 00:04:58.800
And Alisa.
00:04:59.519 --> 00:05:05.680
Yeah, no, what I what we're seeing too a lot of is that people are ingesting these so quickly.
00:05:05.839 --> 00:05:08.480
And the right hand and the left hand aren't always talking.
00:05:08.560 --> 00:05:14.720
So we're seeing a lot of overlap and we're seeing a lot of uh thought process.
00:05:14.800 --> 00:05:29.120
And we are, we're gonna kind of walk through, but I think it it's maybe potentially obvious, but incumbent on everyone to really think about what are we gonna add to our procurement questionnaire uh that makes sense, uh right?
00:05:29.279 --> 00:05:33.839
When are we gonna let them not answer some of the questions or areas which may be appropriate?
00:05:33.920 --> 00:05:39.920
What when are we gonna require a more robust uh completion of a questionnaire and a tool?
00:05:40.160 --> 00:05:51.279
And also how much are we gonna push on our business data to really understand the functionality and potentially the cross-functionality that it's gonna be adopting?
00:05:51.600 --> 00:06:02.639
And so where I like to kind of start to is on square one, which is knowing that everybody is in buy mode right now, buy, bye, buy, buy, peace, peace, peace, right?
00:06:02.720 --> 00:06:05.439
It's also kind of like you deal with your kids, right?
00:06:05.519 --> 00:06:11.680
Like, okay, let's figure out what we need short term, what do we need long term, right?
00:06:11.839 --> 00:06:22.000
And and walking through that mentally enough to know what that our end goal is going to be accomplished, and articulating that end goal.
00:06:23.279 --> 00:06:24.399
Yeah, I think that's a great point.
00:06:24.560 --> 00:06:30.240
And and touching on some of the things that Betsy said, it's it's a lot more than just buying a piece of software.
00:06:30.399 --> 00:06:34.800
It's a lot more than being and buying a piece of IT and it's the new shiny thing.
00:06:34.959 --> 00:06:40.319
And to what you were saying, Alisa, I mean, there's a lot of there's a lot of interest in getting some of these really cool new tools.
00:06:40.480 --> 00:06:43.600
But at the same time, you know, is it really a reimbursement opportunity?
00:06:43.759 --> 00:06:45.439
Is it really a risk transfer?
00:06:45.839 --> 00:06:50.480
A lot of the reimbursement is is just not there yet for some of these technology tools.
00:06:50.639 --> 00:07:05.120
And while you may be transferring some risk, reducing risk, because AI can help reduce errors in many cases, um, AI introduces new risks as well, regulatory risks or risks related to not having a human in the loop when you really should.
00:07:05.199 --> 00:07:08.879
Um, so there's some risk, which I think we're gonna talk about a little bit today.
00:07:09.120 --> 00:07:10.879
But um let me let me turn it over to Betsy.
00:07:10.959 --> 00:07:12.399
Betsy, let me ask you a question.
00:07:12.480 --> 00:07:22.240
Um, before we talk about really how to diligence these deals or how to paper these deals, let's talk about you know what are these companies or what are these buyers actually purchasing?
00:07:22.399 --> 00:07:29.680
Um so tell us a little bit or talk a little bit about what is the RPM, the RTM, the AI vendor landscape look like right now.
00:07:30.240 --> 00:07:31.360
So thanks, Sean.
00:07:31.519 --> 00:07:40.000
Yeah, I think it will be helpful to the rest of the discussion to sort of take a step back and tee up um some of these concepts.
00:07:40.160 --> 00:08:13.439
So first, just as a reminder to everybody, um when we say RPM, generally we're talking about remote patient monitoring, but there are under that two categories, um, one of which is also called RPM or remote physiologic monitoring, um, just to make things confusing, um, you know, which is where you are tracking vital signs, um, you know, automatically recording the body's responses to certain things.
00:08:13.519 --> 00:08:28.720
And then you there's also remote therapeutic monitoring, which is uh monitoring um more patient adherence to their care plan, you know, are they taking medication as prescribed?
00:08:28.879 --> 00:08:33.440
Um are they or aren't they um moving appropriately?
00:08:33.600 --> 00:08:38.000
Are they, you know, are they falling out of bed, not falling out of bed?
00:08:38.159 --> 00:08:55.279
Um and then we're also seeing patient engagement tools which can um encourage patients to collect data um to monitor their frequent frequently chronic conditions.
00:08:55.600 --> 00:09:16.559
Um and these tools, uh the RPM, the remote patient monitoring tools, um, offer a way to try to get ahead of patients who have chronic conditions to reduce the need for them to spend time in the hospital.
00:09:16.799 --> 00:09:23.840
In some cases, they may be able to be discharged home sooner with these uh remote monitoring technologies.
00:09:24.080 --> 00:09:38.559
Um, hopefully delivering better patient outcomes because in real time you're monitoring um someone's insulin is spiking, is you know, um are they um having an arrhythmia?
00:09:38.720 --> 00:09:47.279
And you can address those issues in real time, hopefully preventing a trip to the emergency room or a long and costly hospital stay.
00:09:48.639 --> 00:10:03.039
Um and we're seeing a greater push to use these remote tools because uh hospitals' health systems you know don't have the staffing to keep folks in the hospital.
00:10:03.440 --> 00:10:09.919
Patients don't want to be in the hospital any longer than necessary and frequently do better at home.
00:10:10.159 --> 00:10:17.600
Um and um as we mentioned, there may be some reimbursement um incentive there.
00:10:18.080 --> 00:10:28.480
Um and as you know, and as we've talked about before, we're seeing rapid proliferation with these digital health tools.
00:10:28.639 --> 00:10:35.759
Um, you know, there um there is more to buy than you could money you could spend on it.
00:10:35.919 --> 00:10:40.480
Um and um, you know, we're seeing a lot of functional overlap with these tools.
00:10:40.720 --> 00:10:48.480
Um, we're also seeing um this technology being marketed as a services stack.
00:10:48.639 --> 00:10:56.480
So it's not just you're buying a device, you are buying an entire services um stack.
00:10:56.799 --> 00:11:14.879
Um again, a lot of these vendors are now incorporating AI into their tool, their products, um, you know, whether it's predictive analytics, clinical decision support, workflow automation, or uh patient engagement chatbots.
00:11:15.200 --> 00:11:18.480
Um, you know, AI has a role to play in that.
00:11:18.639 --> 00:11:45.759
And I think as you're looking at tools to purchase, it's important to understand whether the platform you're considering is an AI-enabled platform, meaning one where it's a pre-existing platform, didn't have AI before, but AI is sort of bolted onto the existing platform to add value to um to the product.
00:11:46.399 --> 00:11:48.080
That's one type of platform.
00:11:48.480 --> 00:11:56.960
It's a second platform um that is becoming uh more and more prevalent is the AI-dependent platform.
00:11:57.039 --> 00:12:02.320
And those are platforms where AI is built in from the beginning.
00:12:02.480 --> 00:12:12.240
The AI is critical, it's fundamental to the um services that the platform delivers.
00:12:12.639 --> 00:12:16.559
If you take away the AI, that platform has no value.
00:12:16.799 --> 00:12:29.360
Um, so it's important as you're going through the contracting process, and I think Elisa and Sean will touch on this later, to understand which type of platform you're getting because they have different risks.
00:12:29.840 --> 00:12:53.120
Um, I another thing quickly that I want to touch on with respect to the current landscape is some of the market drivers that we're seeing that are responsible for um the excitement around um all these technologies.
00:12:53.519 --> 00:13:00.159
Um, as I mentioned, provider staffing pressure, um, especially coming out of COVID.
00:13:00.320 --> 00:13:05.600
Healthcare has lost a lot of um, you know, experienced workers.
00:13:05.759 --> 00:13:15.120
And so um this is viewed as a way to expand the workforce without necessarily having to hire additional folks.
00:13:15.360 --> 00:13:26.000
Um, the idea is that remote patient monitoring and these other digital health tools can lead to greater efficiency in your operations.
00:13:26.320 --> 00:13:34.399
Um, they may be able to address some budget pressures, although I would note a lot of these technologies are not cheap to implement.
00:13:34.480 --> 00:13:40.720
Um, so there may be some upfront um financial pain to implement these.
00:13:40.960 --> 00:13:56.960
Um CMS is uh well up in through 2026, CMS has expanded um reimbursement opportunities, and uh we'll touch on that um a little later.
00:13:57.279 --> 00:14:03.840
Um and you know, patients want to receive care at home, nobody wants to be in a hospital.
00:14:03.919 --> 00:14:10.399
So if you can help them um manage their care at home, you know, that's a better alternative.
00:14:10.639 --> 00:14:18.399
And we're also seeing investor pressure on vendors to really commercialize commercialize AI quickly.
00:14:18.639 --> 00:14:24.159
So that is, you know, another driver um in this area.
00:14:24.480 --> 00:14:25.840
So yeah.
00:14:27.360 --> 00:14:28.720
I think the research you want to do.
00:14:28.960 --> 00:14:42.639
I mean, yeah, well, what I was gonna say is I think that's a great tee up really for look the healthcare industry is adopting this because there is some hope for reimbursement here.
00:14:42.879 --> 00:14:49.600
And so, you know, if the the vendors actually also are very smart in making that obvious.
00:14:49.840 --> 00:14:58.159
So if the vendors, you know, selling a reimbursement lift here, you know, Sean, I mean, how much of what the provider is really buying in?
00:14:58.240 --> 00:15:01.039
Is that is that a regulatory risk area?
00:15:01.360 --> 00:15:04.799
And has the ground just you know shifted underneath?
00:15:05.200 --> 00:15:09.039
We're in a highly regulated industry here in healthcare.
00:15:09.200 --> 00:15:14.720
And so that adds just such a new component uh to the matrix.
00:15:15.120 --> 00:15:20.000
Yeah, especially when we're talking about both AI and RPM tools kind of being integrated together.
00:15:20.080 --> 00:15:32.879
There's kind of a regulatory overlay for both of them, and it can lead to a lot of efficiencies, but um but also the way that RPM is reimbursed is is for the treatment management codes, at least, it's reimbursed by the time that you spend.
00:15:33.039 --> 00:15:39.440
And if you incorporate artificial intelligence into it, maybe you needed less time of humans to spend on it.
00:15:39.600 --> 00:15:41.440
And so it's almost counterintuitive.
00:15:41.519 --> 00:15:50.320
We're gonna use AI that costs us a lot of money, but we're gonna actually make less money because we can't bill as much of the codes because we're not having humans you know clocking in and clocking out.
00:15:50.399 --> 00:15:51.519
So it can be a challenge.
00:15:51.600 --> 00:15:56.000
And I think because of that, there are certainly, like you said, what did you say, shifting ground?
00:15:56.159 --> 00:15:59.919
The ground is shifting underneath us possibly for RPM and RTM right now.
00:16:00.080 --> 00:16:03.440
And I you know foresee more evolution in the future.
00:16:03.600 --> 00:16:07.360
And I'll just take a minute to to say what I'm speaking about.
00:16:07.519 --> 00:16:15.679
So the 2027 proposed uh physician fee schedule rule was just released a couple of weeks ago, and we're recording this on uh August 5th of 2026.
00:16:15.759 --> 00:16:17.440
So it was released a couple of weeks ago.
00:16:17.600 --> 00:16:26.639
I think comments are due through mid-September, but one of the main things that they honed in on from a digital health perspective was changes to proposed changes to RPM and RTM.
00:16:26.879 --> 00:16:47.200
And the biggest proposed change, at least from my perspective, is they are proposing to require that clinical staff that is doing the treatment management services, interacting with the patient, and doing all of that incident to a billing provider or supervising provider, that all of those must be direct employees of the provider, which is a huge change.
00:16:47.360 --> 00:16:51.440
It means that essentially outsourced um staffing models can no longer be used.
00:16:51.600 --> 00:17:16.559
And I do think that that is specifically what CMS is targeting, but that would be a major blow to the healthcare industry and especially to those companies that do and use RPM, um, because CMS has repeatedly, since these codes have been available for almost a decade now, said that they can be outsourced and repeatedly emphasized that they can be employed by a third party as long as they are still under the general supervision of that billing provider.
00:17:16.720 --> 00:17:29.279
Um, so that you know is gonna cause a lot of problems if this proposed rule gets finalized for a lot of different companies that are either gonna have to completely restructure their model or go out of business potentially.
00:17:29.440 --> 00:17:34.240
So it's um, you know, companies that are focused on doing that outsourced RPM service.
00:17:34.480 --> 00:17:54.319
Um, it also is gonna potentially cause problems for companies or for healthcare providers that use an MSO structure, you know, an MSO-friendly PC structure where the MSO may employ the clinical staff, or also even practices that don't use an MSO don't outsource it, but they hire 1099's independent contractors to do it instead of uh direct employees.
00:17:54.480 --> 00:17:58.240
So there's a lot, I think, of clarification that CMS needs to do there.
00:17:58.400 --> 00:18:02.400
And also I'm certain that there's gonna be a lot of pushback in the industry.
00:18:02.480 --> 00:18:06.880
So we'll see where the final rule lands in uh in November when it's released.
00:18:07.119 --> 00:18:11.359
But um, but there are some changes on the horizon for RPM and RTM.
00:18:11.519 --> 00:18:18.559
A couple of other ones, I won't go through all of them, but they're also proposing not in this rule, but they've asked for comments on bundling.
00:18:18.640 --> 00:18:24.160
I think there's 17 different codes now for RPM and RTM, bundling those into four codes.
00:18:24.319 --> 00:18:31.599
So that you know, there's a lot more requirements for each of those codes, and it's gonna make things simpler, but it may also lose some opportunities there.
00:18:31.920 --> 00:18:38.319
Um, the other topic I wanted to touch on that is also you know ripe for regulation is artificial intelligence.
00:18:38.640 --> 00:18:44.960
And I will say that AI, a lot of people don't think about this, but AI is already regulated in a lot of different respects.
00:18:45.119 --> 00:18:49.359
The the OIG and Department of Justice for healthcare, they enforce fraud and abuse.
00:18:49.519 --> 00:18:51.359
Those rules apply to AI as well.
00:18:51.599 --> 00:18:58.799
The Office of the National Coordinator for Health IT, they have rules around certified health IT, including those that incorporate AI.
00:18:59.039 --> 00:19:04.000
Um, the FDA regulates medical devices, including software as a medical device.
00:19:04.160 --> 00:19:13.119
Um, and FTC, and this is just a couple of these agencies, but the Federal Trade Commission um regulates unfair and deceptive trade practices and unfair marketing.
00:19:13.200 --> 00:19:18.319
So, you know, you already have a lot of laws that are applicable to how AI should and can be used.
00:19:18.559 --> 00:19:24.480
But what we don't have right now, and there's been a lot of discussion around, is we don't have a comprehensive federal law.
00:19:24.640 --> 00:19:30.480
Um, some states have enacted laws, which makes for a really difficult compliance patchwork.
00:19:30.559 --> 00:19:36.480
Um, and a lot of those states say that healthcare is a sensitive use case, so there's additional guardrails that need to be implemented.
00:19:36.720 --> 00:19:45.440
But thus far, the federal government has really remained hands-off, and that's been certainly been the Trump administration's perspective, at least until very recently.
00:19:45.599 --> 00:19:57.599
But there's a number of executive orders that where the Trump administration has said we're gonna let innovation go just completely unleashed, and we're even gonna try to preempt state laws that are trying to regulate AI.
00:19:57.839 --> 00:20:12.079
Um, just recently, the Trump administration has announced that they're working on an AI framework, and just in the last couple of days, they've said they've not released an AI framework, but a sort of a proposed AI regulatory framework, and they're meeting with AI developers.
00:20:12.240 --> 00:20:15.519
So there's perhaps more uh more in the works there.
00:20:15.680 --> 00:20:28.559
But you know, as of today, we don't have any comprehensive federal regulatory overlay other than those specific agencies and laws that already touched on, um, you know, like FDA, FTC, Department of Justice, et cetera.
00:20:28.880 --> 00:20:31.920
So enough from me, but let me let me turn it over to you guys.
00:20:32.160 --> 00:20:44.960
Any other thoughts on any other sort of areas of legal risk or or even regulatory diligence that we should be looking at for companies that we're partnering with or investing in that are facing these sort of advanced healthcare technology issues?
00:20:45.359 --> 00:20:55.359
Well, I do want to jump off of one of the points that you make, Sean, especially as we think about these laws, whether it's state, federal, et cetera, right?
00:20:55.440 --> 00:21:00.400
Um, and that is not just where we are, but where we're going to be.
00:21:00.880 --> 00:21:24.079
And and potentially anticipating some of that now as um organizations still struggle with what is my governance program going to include, what it is not going to include, what it is going to look like, and how are we going to have to change things in uh five years if there is a federal policy?
00:21:24.240 --> 00:21:30.640
How do we retrofit what we've done and how do we characterize what we're doing now?
00:21:30.880 --> 00:21:59.039
And so that is just one um uh uh push for everyone who's listening to really try to uh spend a few minutes in your respective governance committee meetings or conversations that you're having to try to at least a little bit anticipate some of that future again, depending on how the rules change in November, it could be more immediate or we could be delaying the pain for two, four, six years.
00:21:59.440 --> 00:21:59.759
I don't know.
00:22:00.400 --> 00:22:25.920
It's challenging because to further tease out what Elisa is saying, you know, in December 2025, HHS issued a request for information about accelerating the adoption of AI for clinical care, not just back office support.
00:22:26.559 --> 00:22:44.559
And part of that was soliciting feedback on what payment policy changes are needed to drive innovation to promote access to high-value clinical interventions using AE.
00:22:45.839 --> 00:22:59.839
And it's interesting to contrast that with what Sean talked through in the proposed rule about perhaps scaling back reimbursement for RPM.
00:23:00.000 --> 00:23:27.599
And so it will be interesting to see how this plays out and what policies come out to make it easier for organizations to make the financial commitment to either invest in developing new technology or buying that technology and then implementing it to provide care to their patients.
00:23:27.759 --> 00:23:43.279
So it's um yes, having long, hard discussions about what your governance um program is going to look like and how you're going to structure it is um yes, important.
00:23:44.000 --> 00:23:48.960
As we know, because the winds can change in you know a year, two years, three years.
00:23:49.119 --> 00:23:54.799
Um, you know, um, so it's challenging to say the least.
00:23:55.920 --> 00:23:56.400
Yeah.
00:23:56.720 --> 00:24:01.599
Well, I'm going to use that to jump off of something that I'm interested in talking a little bit about today.
00:24:01.839 --> 00:24:10.079
And that is, you know, um how we're helping the clients, all of us, uh, negotiate these agreements.
00:24:10.240 --> 00:24:37.039
And I I want to start with one thought that I had earlier today as I was reviewing a contract, and that is we've previously had uh software, software lawyers outside of healthcare writing contracts that we've had to ingest and adopt and put in our cute little healthcare things that we have to put in because we are in a highly regulated industry.
00:24:37.519 --> 00:24:41.920
Um very important concepts, thought process, right?
00:24:42.160 --> 00:24:45.599
Uh any kickback concerns, etc.
00:24:46.400 --> 00:25:03.200
I'm seeing that at such an exponentially faster pace now, where um those considerations are absolutely absent from the agreement, even if it is theoretically being sold in healthcare.
00:25:03.359 --> 00:25:14.880
So I have a healthcare provider agreement from an AI vendor that I am looking at, and it was so clearly written by somebody who's never worked in healthcare before.
00:25:15.119 --> 00:25:16.799
And that's difficult, right?
00:25:16.880 --> 00:25:21.440
Because we have client expectations of I needed this yesterday, right?
00:25:21.599 --> 00:25:27.200
And what needs to be in here from a healthcare perspective and where those lines are.
00:25:27.359 --> 00:25:48.559
So fundamentally, first, I will tell you that it is really important as always to put on an incredibly critical eye and not assume that that vendor who purports to be in healthcare may or may not have what it needs in its contract to protect your client.
00:25:48.720 --> 00:25:51.359
Okay, assuming you're on the provider side.
00:25:51.440 --> 00:25:57.279
And if you're on the vendor side, please continue to put in what we need to put in now.
00:25:57.599 --> 00:25:58.880
What are we seeing, right?
00:25:58.960 --> 00:26:01.680
Where are the where are the deals stalling?
00:26:01.839 --> 00:26:07.759
It is also about uh who your client is and and are they now PEO?
00:26:08.240 --> 00:26:25.039
So the expectation is to be fast and quick and uh and and agile as it relates to some of those risk-based areas that traditionally providers have been much more cautious on, right?
00:26:25.279 --> 00:26:48.480
So the shift of risk from a provider, almost in that you know, AWS cloud computing like uh standpoint of we will accept no liability beyond one time, five times fees paid annualized to a uh a provider entity that's like, well, wait a minute, you caused our issue.
00:26:48.960 --> 00:27:00.079
We're looking to you for the expertise, you know, uh short of making us whole a five time fees paid uh is not going to work.
00:27:00.880 --> 00:27:23.519
So uh from the outset, on all of these uh agreements, even more than we've ever had before in our world of um software development, is really setting expectations early on of risk shifting and those traditional indemnification, limitation of liability clauses.
00:27:23.759 --> 00:27:37.599
I'll also note too, funny enough, in the contract that I was looking at this morning, we had four different indemnification clauses: MSA, scope of work, BAA, and then an add-on, right?
00:27:38.000 --> 00:27:47.200
So we're having to make sure that there's real understanding and consistency of the issues.
00:27:47.599 --> 00:28:07.359
And and um in a contract that's only worth$10,000 a year, it is hard to spend uh five to$10,000 on a negotiation, other than the fact that you have four indemnification clauses in this contract that we want to make sure that uh uh have a synergy.
00:28:07.519 --> 00:28:17.119
These are not new issues to AI or RPM or any of these agreements, but they do seem to be coming at a much more difficult clip now.
00:28:17.359 --> 00:28:20.240
What are some of the other things that we're seeing, right?
00:28:21.039 --> 00:28:41.279
Um, and anticipating some of those changes, we are always uh thoughtful and have been certainly since HIPAA came into play as we think about uh derivative data, secondary uses of data, de-identification of data, what does de-identification of data actually mean, right?
00:28:41.440 --> 00:28:54.480
Um, how do we feel about uh information that may be de-identified as defined by the law, but possibly re-identifiable in a quantum computing um type world?
00:28:54.720 --> 00:29:22.799
So I I really want to make sure that as um uh everyone listening today thinks about uh data and ownership and licensing is goes back to what uh Betsy and I were hammering on a few minutes ago, which is making sure that your client is thoughtful and strategic about how they feel about their data.
00:29:23.279 --> 00:29:30.160
And if they're willing to have that de-identified data set no longer owned by them, right?
00:29:30.480 --> 00:29:37.440
That is a decision point that is uh currently likely permissible in the way that it's written.
00:29:38.160 --> 00:29:40.880
Um, but is that what we want?
00:29:41.440 --> 00:30:29.039
Maybe, maybe not, but really making sure that you're clear on what that is is really important, and that's where also we're seeing a lot of uh commingling of platforms and data across these platforms and these analytic uh engines, and so being very clear too, when you're looking at a contract, negotiating it where that data might go outside of that contract, where it may be shared outside of that contract, you should be asking, and uh you should be trying to be one step ahead sometimes uh uh uh of your clients where they may not know or anticipate uh where the other places could be.
00:30:29.279 --> 00:30:38.559
So that de-identification and data ownership uh going beyond the four corners is is, I think, pretty important.
00:30:39.119 --> 00:30:39.599
Right.
00:30:39.920 --> 00:31:00.799
I will also say uh it is important in my estimation of a fairly good clause or clauses in your tool shed um to think about uh uh you know the evaluation of bias and uh safety and ethical concerns ready.
00:31:01.039 --> 00:31:19.599
Um again, it you're seeing more of those coming in the first round of contracts uh to be reviewed, which is great, but we still see it missing from the vast majority of agreements that at least I'm being asked to review today.
00:31:19.839 --> 00:31:41.759
And so making sure that you um have an understanding of where you want to have those clauses and how they will continue through the life of that evergreen agreement, um throughout the the the at least the coming years uh while you're responsible for it, right?
00:31:42.000 --> 00:31:54.640
Um the equity and fairness arguments uh as I'm negotiating, and of course, another war story, I was negotiating a business associate agreement uh about two or two days ago.
00:31:54.960 --> 00:32:08.079
And um as we were looking at the equity and fairness, it was very clear that I was arguing against an AI um tool and not an actual person on the other line, which is fine.
00:32:08.640 --> 00:32:15.279
But uh the AI had not caught itself in talking essentially on both sides of its mouth, right?
00:32:15.440 --> 00:32:19.599
One is this is market, and so we think you should agree to this.
00:32:20.240 --> 00:32:25.920
Um and the other was, well, this is what the only what the rule says, and that's all we're gonna agree to, right?
00:32:26.079 --> 00:32:29.119
And so it was toggling between both those arguments.
00:32:29.279 --> 00:32:33.759
There was a sometimes they can coexist, sometimes they can't, suffices to say.
00:32:33.839 --> 00:32:36.720
And this one, they kind of caught themselves in it.
00:32:36.880 --> 00:32:44.799
Um, it also missed a cross-reference um to an indemnification clause it tried to remove as it was in negotiating another part of the agreement.
00:32:44.960 --> 00:32:56.720
So, you know, it's important to be thinking again about uh as we do, equity and fairness of who's most equipped to be dealing with some of these issues.
00:32:57.119 --> 00:33:09.759
Maybe it's the provider and you want control, but again, as we think about indemnification, bigger companies, uh, who's at risk, thinking about what those might be.
00:33:10.000 --> 00:33:17.440
And that's also where we uh get into some uh called debates on what the reps and warranties are going to look like.
00:33:17.680 --> 00:33:37.519
What will the um uh vendor rep to uh as it's talking about what it can say about how it will not discriminate or how it will keep up with ensuring that its uh tool is not operating in a discriminatory um uh mechanism.
00:33:38.799 --> 00:33:41.599
The privacy and security issues continue.
00:33:41.759 --> 00:34:03.200
They are just uh more complicated, and the risk now is amplified because as we have more and more data and more and more data um uh being replicated through the AI tools, more and more data being utilized for other purposes to hopefully make medical advances, right?
00:34:03.279 --> 00:34:13.920
And and and care advances, but that certainly is going to add to the risk allocation um that that we see uh quite a bit.
00:34:14.480 --> 00:34:31.519
And so the last thing I'll note too is is the complexity that can come with aligning um SLAs uh with the clinical um uh workflows uh and the reimbursement rules.
00:34:32.000 --> 00:34:38.559
Uh reimbursement uh rules in this, uh as we talked about earlier, are quite tricky.
00:34:38.960 --> 00:34:47.199
And um frankly, CMS has said we're looking for this and we we will be paying attention to this.
00:34:48.000 --> 00:34:58.480
So uh I think it's incumbent on us as we're looking at these agreements to be uh anticipating uh where that could go off the rails.
00:34:58.639 --> 00:35:20.800
Um and again, that may come back to a negotiation standpoint, uh, but we also need to be advising our uh business teams that no matter what, some of this you can't shift that risk, and it is going to be born uh to the team because that human in the loop, as Sean mentioned earlier, is hugely important.
00:35:23.119 --> 00:35:32.000
Yeah, I mean, I think that that's a great, it's it's probably not um all-encompassing, but those are those are some serious things you need to be thinking about when looking at these types of contracts.
00:35:32.079 --> 00:35:42.719
And I've experienced the same thing that a lot of these companies are developed by maybe by IT guys, you know, software guys, and they don't really have a lot of experience in healthcare.
00:35:42.800 --> 00:35:46.960
And their lawyers, if they if they're even using lawyers, their lawyers don't have a lot of experience in healthcare.
00:35:47.039 --> 00:35:52.719
So they give you a simple software contract, but it's a lot more than that, and there's a lot more things that need to be considered.
00:35:52.880 --> 00:36:03.119
And the data use and indemnification, those are all really key provisions that are very different in the healthcare context than if it's some other, you know, some other AI application.
00:36:03.280 --> 00:36:04.480
So it's a great point.
00:36:04.719 --> 00:36:09.039
I want to shift gears a little bit and and ask um ask Betsy about the future.
00:36:09.280 --> 00:36:17.440
I mean, Betsy, look forward for us and what what what should we be building into deals now for where this is headed over the next several years?
00:36:18.320 --> 00:36:27.599
Well, let me get out my crystal ball and um uh talk about a few things.
00:36:27.679 --> 00:36:49.599
Um, I think we can anticipate more regulatory development going forward, um, even though we're, you know, the current administration, I think initially a lot of people probably thought that there would be more of a deregulatory atmosphere.
00:36:49.840 --> 00:36:53.760
Um that's not really the case with healthcare.
00:36:54.000 --> 00:37:22.400
Um, so I think um, you know, it we need to be mindful um of some of those developments, obviously monitoring what is happening on with AI regulatory developments, whether there will ever be a federal AI um law, um, but also monitoring you know state AI developments.
00:37:22.480 --> 00:37:29.119
Um, you can't ignore those, and the states are you know aggressively enforcing them.
00:37:29.360 --> 00:37:39.840
Um, you know, even thinking about things like is the tool you're purchasing, could it conceivably be practicing medicine?
00:37:39.920 --> 00:37:52.719
You know, we've had the first case in uh Pennsylvania where um an AI tool has uh is alleged to have um engaged in practicing medicine.
00:37:52.880 --> 00:37:58.800
Um something, you know, that's probably something we didn't think we'd have to worry about a couple years ago.
00:37:59.039 --> 00:38:10.719
Um, you know, I we're also um, you know, you need to think about antitrust scrutiny from the FTC and DOJ.
00:38:10.880 --> 00:38:14.960
You know, a lot of the issues that um Elisa ticked through earlier.
00:38:15.119 --> 00:38:21.519
Um, you know, the FTC has a healthcare task force, so you need to be mindful if you're on the vendor side.
00:38:21.760 --> 00:38:28.079
Is your tool, is your AI doing what you you're saying it's doing?
00:38:28.320 --> 00:38:39.599
Um, you know, um, or are it, you know, could it be are your um representations about it potentially deceptive, you know, untrue?
00:38:39.840 --> 00:39:03.920
Um also increased healthcare fraud, waste, and abuse enforcement at both the federal and the state level, um, you know, which is a little surprising in the current environment, but um, you know, HHS has made it very clear that they are HHS and DOJ, that they are targeting fraud, waste, and abuse.
00:39:04.079 --> 00:39:33.840
And um, you know, while some of these AI products can deliver operational efficiency, um, they may be able to do other things at scale that could, you know, land um healthcare systems and providers um on the wrong side of um you know the OIG or CMS if you're not watching how these what these tools are doing and how they're being used.
00:39:34.079 --> 00:39:37.599
Um state oversight and private equity ownership.
00:39:37.679 --> 00:39:47.599
Um, you know, Elisa had mentioned uh PE ownership before and how that changes some of the risk um calculations when you're contracting.
00:39:47.760 --> 00:39:53.920
Um, you know, but we also expect um, you know, I think that's something to consider.
00:39:54.159 --> 00:39:56.000
Corporate practice of medicine.
00:39:56.079 --> 00:40:06.239
Uh, we're seeing a lot of there's going to continue to be a lot of activity in that area, and you know, you need to be mindful of that as you're contracting.
00:40:06.480 --> 00:40:31.199
Um, you know, FDA oversight, you know, um things you need to, especially with respect to AI-enabled medical devices, um state oversight of healthcare transactions, um, you know, um, and you know, this patchwork of state privacy laws, I think that's going to continue.
00:40:31.599 --> 00:40:43.280
Um especially as and we see more states developing or passing um healthcare data specific privacy laws.
00:40:43.519 --> 00:40:49.280
Um I mentioned the state AI laws, you know, healthcare licensing laws.
00:40:49.519 --> 00:41:01.440
Um, so there is a lot of balls in the air potentially, um, you know, which is going to make contracting um, I think, um, more challenging.
00:41:01.599 --> 00:41:05.119
Um, you know, a lot more things that need to be addressed.
00:41:05.440 --> 00:41:10.239
Also stepping um back a little more.
00:41:10.400 --> 00:41:15.519
Um, I think we're going to see more AI ready contract templates.
00:41:15.760 --> 00:41:21.599
So moving from these point solutions to platform contracting.
00:41:21.760 --> 00:41:28.239
Again, it's, you know, you're not just buying a device or software, you are getting a platform.
00:41:28.639 --> 00:41:32.719
Um, and um I think Elisa mentioned this earlier.
00:41:33.039 --> 00:41:37.119
Clients need to be ready for shorter innovation cycles.
00:41:37.360 --> 00:41:46.559
So um, you know, your has typically you may have done, you know, looked at a contract every three to five years.
00:41:46.880 --> 00:41:58.000
That cycle is going to shorten significantly with the rapidity with which new tools are developing and being improved.
00:41:58.239 --> 00:42:06.800
Um, continuous compliance obligations, um, and you know, AI is here to stay.
00:42:06.960 --> 00:42:10.960
It is going to be a permanent fixture in these agreements.
00:42:11.039 --> 00:42:19.920
Um, and so we all have to learn you know how to um live with that and help our clients adjust to that.
00:42:20.800 --> 00:42:30.000
So, but uh Alisa and Sean, um, you know, anything else that you see in your crystal balls uh going forward.
00:42:31.280 --> 00:42:43.760
I mean, I'll say that I I do think that we'll ex we should expect to see some federal AI regulation, whether that be in the Trump administration doing something informally or maybe regulatory.
00:42:43.920 --> 00:42:53.119
Um, and I don't think that the Congress Congress is gonna take it up anytime soon, but I think it's potentially over the next couple of years after these midterms that are coming up this fall.
00:42:53.199 --> 00:42:56.800
Um, so I do see foresee some sort of activity there.
00:42:58.639 --> 00:43:00.559
Gosh, I wish I were that optimistic.
00:43:00.719 --> 00:43:04.320
I'm gonna take your thing and and um I I I like that.
00:43:04.400 --> 00:43:08.719
I mean, I really do hope you're right on that because I do think it will help.
00:43:08.880 --> 00:43:17.440
Uh, you know, as much as a lot of us don't necessarily love over-regulation, some of that clarity I think would would go a long way.
00:43:17.679 --> 00:43:18.000
Yes.
00:43:18.239 --> 00:43:26.559
But um, you know, Sean, I mean, as we wrap up, do you have any key takeaways you think we should make sure we can circle back on?
00:43:27.119 --> 00:43:30.559
Yeah, I mean, I'll just wrap it up, I guess, with kind of how I started.
00:43:30.719 --> 00:43:40.559
And that is that these AI and RPM, RTM, these these contracts, these deals, these digital health solutions are not just IT contracts.
00:43:40.639 --> 00:43:47.199
There's more that needs to be thought of when diligencing these solutions and contracting with them and then ultimately using them.
00:43:47.280 --> 00:43:50.000
Um, but it's not it's not all doom and gloom.
00:43:50.159 --> 00:43:59.199
I mean, these there's just some additional considerations that we really need to think about and really think about these IT solutions with that healthcare overlay that you were talking about.
00:43:59.440 --> 00:44:18.639
And you know, think about these risks, think about the regulatory impact, think about the operational impact, think about the reimbursement impact, and and it just should be evaluated and assessed, and then ultimately that risk allocated between the parties and maybe considered in the contracting, including in the pricing.
00:44:18.800 --> 00:44:31.920
You know, you mentioned earlier, Lisa, that there that granting the use of de-identified data is something that is typically requested and oftentimes demanded by an AI vendor, but that has real value.
00:44:32.079 --> 00:44:35.679
So, how are we going to build that into the pricing for this deal?
00:44:35.840 --> 00:44:46.880
Um, I think all of those are really important considerations, but just as a takeaway, look, these are these are these are great opportunities and it's a it's a great world that we're living in to be able to access this great technology.
00:44:47.039 --> 00:44:50.960
But these are a lot more than just IT contracts, and there's a lot more to be thinking about.
00:44:51.599 --> 00:44:51.840
Great.
00:44:52.000 --> 00:44:53.039
Thank you, everyone.
00:44:53.360 --> 00:45:07.679
Yes, and I would just say, yeah, please feel free to make comments to the proposed rule that Sean mentioned, um, address on the issue of RPM and reimbursement.
00:45:07.760 --> 00:45:15.519
Um, you know, that that is another takeaway for the audience if they um are inclined to submit comments on that.
00:45:15.679 --> 00:45:18.960
So I feel we would be remiss if we did not mention that.
00:45:19.280 --> 00:45:20.079
Well said.
00:45:20.639 --> 00:45:22.559
Yeah, well, thank you for joining us for this podcast.
00:45:22.719 --> 00:45:24.239
Appreciate it, Betsy and Elisa.
00:45:29.039 --> 00:45:34.960
If you enjoyed this episode, be sure to subscribe to AHLA Speaking of Health Law wherever you get your podcasts.
00:45:35.119 --> 00:45:51.840
For more information about AHLA and the educational resources available to the health law community, visit American Health Law.org and stay updated on breaking healthcare industry news from the major media outlets with AHLA's Health Law Daily Podcast, exclusively for AHLA comprehensive members.
00:45:52.079 --> 00:45:59.039
To subscribe and add this private podcast feed to your podcast at once, go to americanhealthlaw.org slash daily podcast.com.
00:00:04.719 --> 00:00:10.080
This episode of AHLA Speaking of Health Law is brought to you by AHLA members and donors like you.
00:00:10.320 --> 00:00:13.679
For more information, visit AmericanHealthlaw.org.
00:00:17.199 --> 00:00:22.559
So, what are you really buying when you contract for AI-enabled digital health tools?
00:00:22.719 --> 00:00:42.640
This is the question that we're going to be discussing today, and especially in the context of things like AI moving into a more regulated space, the growth of remote patient monitoring and the potential changes to reimbursement in the regulatory atmosphere for RPM, as well as interoperability and patient engagement challenges as well as strategies.
00:00:42.880 --> 00:00:48.320
So one of the biggest things and challenges in the in the industry right now is fragmentation.
00:00:48.399 --> 00:00:58.159
And that's fragmentation in a couple of different ways, EHRs and different data sources and getting those things to talk to talk to each other, as well as fragmentation, I guess, among different states.
00:00:58.320 --> 00:01:03.920
There's so many different state laws that regulate this differently, as well as the federal government kind of as an overlay on top.
00:01:04.079 --> 00:01:26.879
So our job really as lawyers that are advising these companies, that are maybe investing in healthcare technology or partnering with these technology companies, is to turn this regulatory risk, turn these operational risks into diligence questions, into contract terms, into policies and procedures and training for people that are using these tools to ensure it's being done safely and responsibly.
00:01:27.040 --> 00:01:29.040
So that's essentially what we're going to be talking about today.
00:01:29.120 --> 00:01:35.359
I'm Sean Sullivan, and I'm a healthcare regulatory attorney and a partner at All Sinner Bird in Atlanta.
00:01:35.519 --> 00:01:41.120
I am joined today by Elisa Chessler and Betsy Hodge, and welcome to this AHLA podcast.
00:01:41.439 --> 00:01:50.480
I will note that this podcast is based on our session from HLA's 2026 Transactions Conference in Nashville from back in April.
00:01:50.719 --> 00:01:53.040
But let me turn it over to Elisa and Betsy.
00:01:53.120 --> 00:02:04.799
I'd like you each to each to introduce yourselves and I guess you know answer or talk briefly about the question: what are you really buying when you contract for an AI-enabled or some other advanced digital health tool?
00:02:05.200 --> 00:02:06.480
Elisa, do you want to go first?
00:02:07.040 --> 00:02:08.000
Thanks, Sean.
00:02:08.400 --> 00:02:09.599
Happy to be with everyone today.
00:02:09.680 --> 00:02:11.840
My name is Elisa Chester, as mentioned before.
00:02:12.000 --> 00:02:17.599
I'm a shareholder at Baker Donaldson based in Nashville and Washington, DC.
00:02:18.240 --> 00:02:55.360
And working a lot with clients on the adoption of AI tools and thinking about the best way to make sure that we are holistically thinking through not just those short-term issues of how is it working effectively now, but thinking through where are we going to find ourselves in the integration of those tools and where are we going to find ourselves in five years and 10 years with all the information that the provider community is going to be ingesting and uh dealing with.
00:02:58.639 --> 00:03:00.240
And I'm Betsy Hodge.
00:03:00.319 --> 00:03:05.599
I'm a partner at Akerman LLP in our West Palm Beach office.
00:03:05.759 --> 00:03:26.719
I'm in our healthcare practice group, and I focus on um the provider side of um purchasing um and investigating um these RPM tools and how they can improve my clients' uh care and also help their bottom line.
00:03:26.879 --> 00:03:29.439
So what about the what about the question?
00:03:29.520 --> 00:03:44.319
You know, what do what do what should clients be thinking about, or what should our clients, or if you're in-house, what are you thinking about when you're contracting for these AI-enabled tools or these, you know, other advanced digital health tools like remote patient monitoring platforms or interoperable uh software.
00:03:44.479 --> 00:03:45.599
What are you guys thinking?
00:03:46.000 --> 00:03:48.319
You know, it's a it's a software tool.
00:03:48.479 --> 00:04:00.400
And so a lot of times the contracting folks are looking at this as another software um application, but it's you know, RPM is more than that.
00:04:00.560 --> 00:04:08.000
Um, you know, as I mentioned before, you're also looking for operation something that can deliver operational efficiency.
00:04:08.240 --> 00:04:12.400
Um, and um Sean, you and Lisa touched on this too.
00:04:12.560 --> 00:04:32.399
Also um getting clinical insights and decision support from all the data that is being collected, and which then creates some issues around how do you safeguard that data and you know um responsibly use that data.
00:04:32.560 --> 00:04:36.720
Um, you know, and you also have some reimbursement opportunities.
00:04:36.800 --> 00:04:40.879
We'll touch on that, I think, a little bit later in the conversation.
00:04:41.360 --> 00:04:44.560
Um and risk transfer.
00:04:45.120 --> 00:04:57.680
Um I think so I think those are um some of the things that you're buying when you're buying um when you contract for you know AI enabled digital health.
00:04:57.759 --> 00:04:58.800
And Alisa.
00:04:59.519 --> 00:05:05.680
Yeah, no, what I what we're seeing too a lot of is that people are ingesting these so quickly.
00:05:05.839 --> 00:05:08.480
And the right hand and the left hand aren't always talking.
00:05:08.560 --> 00:05:14.720
So we're seeing a lot of overlap and we're seeing a lot of uh thought process.
00:05:14.800 --> 00:05:29.120
And we are, we're gonna kind of walk through, but I think it it's maybe potentially obvious, but incumbent on everyone to really think about what are we gonna add to our procurement questionnaire uh that makes sense, uh right?
00:05:29.279 --> 00:05:33.839
When are we gonna let them not answer some of the questions or areas which may be appropriate?
00:05:33.920 --> 00:05:39.920
What when are we gonna require a more robust uh completion of a questionnaire and a tool?
00:05:40.160 --> 00:05:51.279
And also how much are we gonna push on our business data to really understand the functionality and potentially the cross-functionality that it's gonna be adopting?
00:05:51.600 --> 00:06:02.639
And so where I like to kind of start to is on square one, which is knowing that everybody is in buy mode right now, buy, bye, buy, buy, peace, peace, peace, right?
00:06:02.720 --> 00:06:05.439
It's also kind of like you deal with your kids, right?
00:06:05.519 --> 00:06:11.680
Like, okay, let's figure out what we need short term, what do we need long term, right?
00:06:11.839 --> 00:06:22.000
And and walking through that mentally enough to know what that our end goal is going to be accomplished, and articulating that end goal.
00:06:23.279 --> 00:06:24.399
Yeah, I think that's a great point.
00:06:24.560 --> 00:06:30.240
And and touching on some of the things that Betsy said, it's it's a lot more than just buying a piece of software.
00:06:30.399 --> 00:06:34.800
It's a lot more than being and buying a piece of IT and it's the new shiny thing.
00:06:34.959 --> 00:06:40.319
And to what you were saying, Alisa, I mean, there's a lot of there's a lot of interest in getting some of these really cool new tools.
00:06:40.480 --> 00:06:43.600
But at the same time, you know, is it really a reimbursement opportunity?
00:06:43.759 --> 00:06:45.439
Is it really a risk transfer?
00:06:45.839 --> 00:06:50.480
A lot of the reimbursement is is just not there yet for some of these technology tools.
00:06:50.639 --> 00:07:05.120
And while you may be transferring some risk, reducing risk, because AI can help reduce errors in many cases, um, AI introduces new risks as well, regulatory risks or risks related to not having a human in the loop when you really should.
00:07:05.199 --> 00:07:08.879
Um, so there's some risk, which I think we're gonna talk about a little bit today.
00:07:09.120 --> 00:07:10.879
But um let me let me turn it over to Betsy.
00:07:10.959 --> 00:07:12.399
Betsy, let me ask you a question.
00:07:12.480 --> 00:07:22.240
Um, before we talk about really how to diligence these deals or how to paper these deals, let's talk about you know what are these companies or what are these buyers actually purchasing?
00:07:22.399 --> 00:07:29.680
Um so tell us a little bit or talk a little bit about what is the RPM, the RTM, the AI vendor landscape look like right now.
00:07:30.240 --> 00:07:31.360
So thanks, Sean.
00:07:31.519 --> 00:07:40.000
Yeah, I think it will be helpful to the rest of the discussion to sort of take a step back and tee up um some of these concepts.
00:07:40.160 --> 00:08:13.439
So first, just as a reminder to everybody, um when we say RPM, generally we're talking about remote patient monitoring, but there are under that two categories, um, one of which is also called RPM or remote physiologic monitoring, um, just to make things confusing, um, you know, which is where you are tracking vital signs, um, you know, automatically recording the body's responses to certain things.
00:08:13.519 --> 00:08:28.720
And then you there's also remote therapeutic monitoring, which is uh monitoring um more patient adherence to their care plan, you know, are they taking medication as prescribed?
00:08:28.879 --> 00:08:33.440
Um are they or aren't they um moving appropriately?
00:08:33.600 --> 00:08:38.000
Are they, you know, are they falling out of bed, not falling out of bed?
00:08:38.159 --> 00:08:55.279
Um and then we're also seeing patient engagement tools which can um encourage patients to collect data um to monitor their frequent frequently chronic conditions.
00:08:55.600 --> 00:09:16.559
Um and these tools, uh the RPM, the remote patient monitoring tools, um, offer a way to try to get ahead of patients who have chronic conditions to reduce the need for them to spend time in the hospital.
00:09:16.799 --> 00:09:23.840
In some cases, they may be able to be discharged home sooner with these uh remote monitoring technologies.
00:09:24.080 --> 00:09:38.559
Um, hopefully delivering better patient outcomes because in real time you're monitoring um someone's insulin is spiking, is you know, um are they um having an arrhythmia?
00:09:38.720 --> 00:09:47.279
And you can address those issues in real time, hopefully preventing a trip to the emergency room or a long and costly hospital stay.
00:09:48.639 --> 00:10:03.039
Um and we're seeing a greater push to use these remote tools because uh hospitals' health systems you know don't have the staffing to keep folks in the hospital.
00:10:03.440 --> 00:10:09.919
Patients don't want to be in the hospital any longer than necessary and frequently do better at home.
00:10:10.159 --> 00:10:17.600
Um and um as we mentioned, there may be some reimbursement um incentive there.
00:10:18.080 --> 00:10:28.480
Um and as you know, and as we've talked about before, we're seeing rapid proliferation with these digital health tools.
00:10:28.639 --> 00:10:35.759
Um, you know, there um there is more to buy than you could money you could spend on it.
00:10:35.919 --> 00:10:40.480
Um and um, you know, we're seeing a lot of functional overlap with these tools.
00:10:40.720 --> 00:10:48.480
Um, we're also seeing um this technology being marketed as a services stack.
00:10:48.639 --> 00:10:56.480
So it's not just you're buying a device, you are buying an entire services um stack.
00:10:56.799 --> 00:11:14.879
Um again, a lot of these vendors are now incorporating AI into their tool, their products, um, you know, whether it's predictive analytics, clinical decision support, workflow automation, or uh patient engagement chatbots.
00:11:15.200 --> 00:11:18.480
Um, you know, AI has a role to play in that.
00:11:18.639 --> 00:11:45.759
And I think as you're looking at tools to purchase, it's important to understand whether the platform you're considering is an AI-enabled platform, meaning one where it's a pre-existing platform, didn't have AI before, but AI is sort of bolted onto the existing platform to add value to um to the product.
00:11:46.399 --> 00:11:48.080
That's one type of platform.
00:11:48.480 --> 00:11:56.960
It's a second platform um that is becoming uh more and more prevalent is the AI-dependent platform.
00:11:57.039 --> 00:12:02.320
And those are platforms where AI is built in from the beginning.
00:12:02.480 --> 00:12:12.240
The AI is critical, it's fundamental to the um services that the platform delivers.
00:12:12.639 --> 00:12:16.559
If you take away the AI, that platform has no value.
00:12:16.799 --> 00:12:29.360
Um, so it's important as you're going through the contracting process, and I think Elisa and Sean will touch on this later, to understand which type of platform you're getting because they have different risks.
00:12:29.840 --> 00:12:53.120
Um, I another thing quickly that I want to touch on with respect to the current landscape is some of the market drivers that we're seeing that are responsible for um the excitement around um all these technologies.
00:12:53.519 --> 00:13:00.159
Um, as I mentioned, provider staffing pressure, um, especially coming out of COVID.
00:13:00.320 --> 00:13:05.600
Healthcare has lost a lot of um, you know, experienced workers.
00:13:05.759 --> 00:13:15.120
And so um this is viewed as a way to expand the workforce without necessarily having to hire additional folks.
00:13:15.360 --> 00:13:26.000
Um, the idea is that remote patient monitoring and these other digital health tools can lead to greater efficiency in your operations.
00:13:26.320 --> 00:13:34.399
Um, they may be able to address some budget pressures, although I would note a lot of these technologies are not cheap to implement.
00:13:34.480 --> 00:13:40.720
Um, so there may be some upfront um financial pain to implement these.
00:13:40.960 --> 00:13:56.960
Um CMS is uh well up in through 2026, CMS has expanded um reimbursement opportunities, and uh we'll touch on that um a little later.
00:13:57.279 --> 00:14:03.840
Um and you know, patients want to receive care at home, nobody wants to be in a hospital.
00:14:03.919 --> 00:14:10.399
So if you can help them um manage their care at home, you know, that's a better alternative.
00:14:10.639 --> 00:14:18.399
And we're also seeing investor pressure on vendors to really commercialize commercialize AI quickly.
00:14:18.639 --> 00:14:24.159
So that is, you know, another driver um in this area.
00:14:24.480 --> 00:14:25.840
So yeah.
00:14:27.360 --> 00:14:28.720
I think the research you want to do.
00:14:28.960 --> 00:14:42.639
I mean, yeah, well, what I was gonna say is I think that's a great tee up really for look the healthcare industry is adopting this because there is some hope for reimbursement here.
00:14:42.879 --> 00:14:49.600
And so, you know, if the the vendors actually also are very smart in making that obvious.
00:14:49.840 --> 00:14:58.159
So if the vendors, you know, selling a reimbursement lift here, you know, Sean, I mean, how much of what the provider is really buying in?
00:14:58.240 --> 00:15:01.039
Is that is that a regulatory risk area?
00:15:01.360 --> 00:15:04.799
And has the ground just you know shifted underneath?
00:15:05.200 --> 00:15:09.039
We're in a highly regulated industry here in healthcare.
00:15:09.200 --> 00:15:14.720
And so that adds just such a new component uh to the matrix.
00:15:15.120 --> 00:15:20.000
Yeah, especially when we're talking about both AI and RPM tools kind of being integrated together.
00:15:20.080 --> 00:15:32.879
There's kind of a regulatory overlay for both of them, and it can lead to a lot of efficiencies, but um but also the way that RPM is reimbursed is is for the treatment management codes, at least, it's reimbursed by the time that you spend.
00:15:33.039 --> 00:15:39.440
And if you incorporate artificial intelligence into it, maybe you needed less time of humans to spend on it.
00:15:39.600 --> 00:15:41.440
And so it's almost counterintuitive.
00:15:41.519 --> 00:15:50.320
We're gonna use AI that costs us a lot of money, but we're gonna actually make less money because we can't bill as much of the codes because we're not having humans you know clocking in and clocking out.
00:15:50.399 --> 00:15:51.519
So it can be a challenge.
00:15:51.600 --> 00:15:56.000
And I think because of that, there are certainly, like you said, what did you say, shifting ground?
00:15:56.159 --> 00:15:59.919
The ground is shifting underneath us possibly for RPM and RTM right now.
00:16:00.080 --> 00:16:03.440
And I you know foresee more evolution in the future.
00:16:03.600 --> 00:16:07.360
And I'll just take a minute to to say what I'm speaking about.
00:16:07.519 --> 00:16:15.679
So the 2027 proposed uh physician fee schedule rule was just released a couple of weeks ago, and we're recording this on uh August 5th of 2026.
00:16:15.759 --> 00:16:17.440
So it was released a couple of weeks ago.
00:16:17.600 --> 00:16:26.639
I think comments are due through mid-September, but one of the main things that they honed in on from a digital health perspective was changes to proposed changes to RPM and RTM.
00:16:26.879 --> 00:16:47.200
And the biggest proposed change, at least from my perspective, is they are proposing to require that clinical staff that is doing the treatment management services, interacting with the patient, and doing all of that incident to a billing provider or supervising provider, that all of those must be direct employees of the provider, which is a huge change.
00:16:47.360 --> 00:16:51.440
It means that essentially outsourced um staffing models can no longer be used.
00:16:51.600 --> 00:17:16.559
And I do think that that is specifically what CMS is targeting, but that would be a major blow to the healthcare industry and especially to those companies that do and use RPM, um, because CMS has repeatedly, since these codes have been available for almost a decade now, said that they can be outsourced and repeatedly emphasized that they can be employed by a third party as long as they are still under the general supervision of that billing provider.
00:17:16.720 --> 00:17:29.279
Um, so that you know is gonna cause a lot of problems if this proposed rule gets finalized for a lot of different companies that are either gonna have to completely restructure their model or go out of business potentially.
00:17:29.440 --> 00:17:34.240
So it's um, you know, companies that are focused on doing that outsourced RPM service.
00:17:34.480 --> 00:17:54.319
Um, it also is gonna potentially cause problems for companies or for healthcare providers that use an MSO structure, you know, an MSO-friendly PC structure where the MSO may employ the clinical staff, or also even practices that don't use an MSO don't outsource it, but they hire 1099's independent contractors to do it instead of uh direct employees.
00:17:54.480 --> 00:17:58.240
So there's a lot, I think, of clarification that CMS needs to do there.
00:17:58.400 --> 00:18:02.400
And also I'm certain that there's gonna be a lot of pushback in the industry.
00:18:02.480 --> 00:18:06.880
So we'll see where the final rule lands in uh in November when it's released.
00:18:07.119 --> 00:18:11.359
But um, but there are some changes on the horizon for RPM and RTM.
00:18:11.519 --> 00:18:18.559
A couple of other ones, I won't go through all of them, but they're also proposing not in this rule, but they've asked for comments on bundling.
00:18:18.640 --> 00:18:24.160
I think there's 17 different codes now for RPM and RTM, bundling those into four codes.
00:18:24.319 --> 00:18:31.599
So that you know, there's a lot more requirements for each of those codes, and it's gonna make things simpler, but it may also lose some opportunities there.
00:18:31.920 --> 00:18:38.319
Um, the other topic I wanted to touch on that is also you know ripe for regulation is artificial intelligence.
00:18:38.640 --> 00:18:44.960
And I will say that AI, a lot of people don't think about this, but AI is already regulated in a lot of different respects.
00:18:45.119 --> 00:18:49.359
The the OIG and Department of Justice for healthcare, they enforce fraud and abuse.
00:18:49.519 --> 00:18:51.359
Those rules apply to AI as well.
00:18:51.599 --> 00:18:58.799
The Office of the National Coordinator for Health IT, they have rules around certified health IT, including those that incorporate AI.
00:18:59.039 --> 00:19:04.000
Um, the FDA regulates medical devices, including software as a medical device.
00:19:04.160 --> 00:19:13.119
Um, and FTC, and this is just a couple of these agencies, but the Federal Trade Commission um regulates unfair and deceptive trade practices and unfair marketing.
00:19:13.200 --> 00:19:18.319
So, you know, you already have a lot of laws that are applicable to how AI should and can be used.
00:19:18.559 --> 00:19:24.480
But what we don't have right now, and there's been a lot of discussion around, is we don't have a comprehensive federal law.
00:19:24.640 --> 00:19:30.480
Um, some states have enacted laws, which makes for a really difficult compliance patchwork.
00:19:30.559 --> 00:19:36.480
Um, and a lot of those states say that healthcare is a sensitive use case, so there's additional guardrails that need to be implemented.
00:19:36.720 --> 00:19:45.440
But thus far, the federal government has really remained hands-off, and that's been certainly been the Trump administration's perspective, at least until very recently.
00:19:45.599 --> 00:19:57.599
But there's a number of executive orders that where the Trump administration has said we're gonna let innovation go just completely unleashed, and we're even gonna try to preempt state laws that are trying to regulate AI.
00:19:57.839 --> 00:20:12.079
Um, just recently, the Trump administration has announced that they're working on an AI framework, and just in the last couple of days, they've said they've not released an AI framework, but a sort of a proposed AI regulatory framework, and they're meeting with AI developers.
00:20:12.240 --> 00:20:15.519
So there's perhaps more uh more in the works there.
00:20:15.680 --> 00:20:28.559
But you know, as of today, we don't have any comprehensive federal regulatory overlay other than those specific agencies and laws that already touched on, um, you know, like FDA, FTC, Department of Justice, et cetera.
00:20:28.880 --> 00:20:31.920
So enough from me, but let me let me turn it over to you guys.
00:20:32.160 --> 00:20:44.960
Any other thoughts on any other sort of areas of legal risk or or even regulatory diligence that we should be looking at for companies that we're partnering with or investing in that are facing these sort of advanced healthcare technology issues?
00:20:45.359 --> 00:20:55.359
Well, I do want to jump off of one of the points that you make, Sean, especially as we think about these laws, whether it's state, federal, et cetera, right?
00:20:55.440 --> 00:21:00.400
Um, and that is not just where we are, but where we're going to be.
00:21:00.880 --> 00:21:24.079
And and potentially anticipating some of that now as um organizations still struggle with what is my governance program going to include, what it is not going to include, what it is going to look like, and how are we going to have to change things in uh five years if there is a federal policy?
00:21:24.240 --> 00:21:30.640
How do we retrofit what we've done and how do we characterize what we're doing now?
00:21:30.880 --> 00:21:59.039
And so that is just one um uh uh push for everyone who's listening to really try to uh spend a few minutes in your respective governance committee meetings or conversations that you're having to try to at least a little bit anticipate some of that future again, depending on how the rules change in November, it could be more immediate or we could be delaying the pain for two, four, six years.
00:21:59.440 --> 00:21:59.759
I don't know.
00:22:00.400 --> 00:22:25.920
It's challenging because to further tease out what Elisa is saying, you know, in December 2025, HHS issued a request for information about accelerating the adoption of AI for clinical care, not just back office support.
00:22:26.559 --> 00:22:44.559
And part of that was soliciting feedback on what payment policy changes are needed to drive innovation to promote access to high-value clinical interventions using AE.
00:22:45.839 --> 00:22:59.839
And it's interesting to contrast that with what Sean talked through in the proposed rule about perhaps scaling back reimbursement for RPM.
00:23:00.000 --> 00:23:27.599
And so it will be interesting to see how this plays out and what policies come out to make it easier for organizations to make the financial commitment to either invest in developing new technology or buying that technology and then implementing it to provide care to their patients.
00:23:27.759 --> 00:23:43.279
So it's um yes, having long, hard discussions about what your governance um program is going to look like and how you're going to structure it is um yes, important.
00:23:44.000 --> 00:23:48.960
As we know, because the winds can change in you know a year, two years, three years.
00:23:49.119 --> 00:23:54.799
Um, you know, um, so it's challenging to say the least.
00:23:55.920 --> 00:23:56.400
Yeah.
00:23:56.720 --> 00:24:01.599
Well, I'm going to use that to jump off of something that I'm interested in talking a little bit about today.
00:24:01.839 --> 00:24:10.079
And that is, you know, um how we're helping the clients, all of us, uh, negotiate these agreements.
00:24:10.240 --> 00:24:37.039
And I I want to start with one thought that I had earlier today as I was reviewing a contract, and that is we've previously had uh software, software lawyers outside of healthcare writing contracts that we've had to ingest and adopt and put in our cute little healthcare things that we have to put in because we are in a highly regulated industry.
00:24:37.519 --> 00:24:41.920
Um very important concepts, thought process, right?
00:24:42.160 --> 00:24:45.599
Uh any kickback concerns, etc.
00:24:46.400 --> 00:25:03.200
I'm seeing that at such an exponentially faster pace now, where um those considerations are absolutely absent from the agreement, even if it is theoretically being sold in healthcare.
00:25:03.359 --> 00:25:14.880
So I have a healthcare provider agreement from an AI vendor that I am looking at, and it was so clearly written by somebody who's never worked in healthcare before.
00:25:15.119 --> 00:25:16.799
And that's difficult, right?
00:25:16.880 --> 00:25:21.440
Because we have client expectations of I needed this yesterday, right?
00:25:21.599 --> 00:25:27.200
And what needs to be in here from a healthcare perspective and where those lines are.
00:25:27.359 --> 00:25:48.559
So fundamentally, first, I will tell you that it is really important as always to put on an incredibly critical eye and not assume that that vendor who purports to be in healthcare may or may not have what it needs in its contract to protect your client.
00:25:48.720 --> 00:25:51.359
Okay, assuming you're on the provider side.
00:25:51.440 --> 00:25:57.279
And if you're on the vendor side, please continue to put in what we need to put in now.
00:25:57.599 --> 00:25:58.880
What are we seeing, right?
00:25:58.960 --> 00:26:01.680
Where are the where are the deals stalling?
00:26:01.839 --> 00:26:07.759
It is also about uh who your client is and and are they now PEO?
00:26:08.240 --> 00:26:25.039
So the expectation is to be fast and quick and uh and and agile as it relates to some of those risk-based areas that traditionally providers have been much more cautious on, right?
00:26:25.279 --> 00:26:48.480
So the shift of risk from a provider, almost in that you know, AWS cloud computing like uh standpoint of we will accept no liability beyond one time, five times fees paid annualized to a uh a provider entity that's like, well, wait a minute, you caused our issue.
00:26:48.960 --> 00:27:00.079
We're looking to you for the expertise, you know, uh short of making us whole a five time fees paid uh is not going to work.
00:27:00.880 --> 00:27:23.519
So uh from the outset, on all of these uh agreements, even more than we've ever had before in our world of um software development, is really setting expectations early on of risk shifting and those traditional indemnification, limitation of liability clauses.
00:27:23.759 --> 00:27:37.599
I'll also note too, funny enough, in the contract that I was looking at this morning, we had four different indemnification clauses: MSA, scope of work, BAA, and then an add-on, right?
00:27:38.000 --> 00:27:47.200
So we're having to make sure that there's real understanding and consistency of the issues.
00:27:47.599 --> 00:28:07.359
And and um in a contract that's only worth$10,000 a year, it is hard to spend uh five to$10,000 on a negotiation, other than the fact that you have four indemnification clauses in this contract that we want to make sure that uh uh have a synergy.
00:28:07.519 --> 00:28:17.119
These are not new issues to AI or RPM or any of these agreements, but they do seem to be coming at a much more difficult clip now.
00:28:17.359 --> 00:28:20.240
What are some of the other things that we're seeing, right?
00:28:21.039 --> 00:28:41.279
Um, and anticipating some of those changes, we are always uh thoughtful and have been certainly since HIPAA came into play as we think about uh derivative data, secondary uses of data, de-identification of data, what does de-identification of data actually mean, right?
00:28:41.440 --> 00:28:54.480
Um, how do we feel about uh information that may be de-identified as defined by the law, but possibly re-identifiable in a quantum computing um type world?
00:28:54.720 --> 00:29:22.799
So I I really want to make sure that as um uh everyone listening today thinks about uh data and ownership and licensing is goes back to what uh Betsy and I were hammering on a few minutes ago, which is making sure that your client is thoughtful and strategic about how they feel about their data.
00:29:23.279 --> 00:29:30.160
And if they're willing to have that de-identified data set no longer owned by them, right?
00:29:30.480 --> 00:29:37.440
That is a decision point that is uh currently likely permissible in the way that it's written.
00:29:38.160 --> 00:29:40.880
Um, but is that what we want?
00:29:41.440 --> 00:30:29.039
Maybe, maybe not, but really making sure that you're clear on what that is is really important, and that's where also we're seeing a lot of uh commingling of platforms and data across these platforms and these analytic uh engines, and so being very clear too, when you're looking at a contract, negotiating it where that data might go outside of that contract, where it may be shared outside of that contract, you should be asking, and uh you should be trying to be one step ahead sometimes uh uh uh of your clients where they may not know or anticipate uh where the other places could be.
00:30:29.279 --> 00:30:38.559
So that de-identification and data ownership uh going beyond the four corners is is, I think, pretty important.
00:30:39.119 --> 00:30:39.599
Right.
00:30:39.920 --> 00:31:00.799
I will also say uh it is important in my estimation of a fairly good clause or clauses in your tool shed um to think about uh uh you know the evaluation of bias and uh safety and ethical concerns ready.
00:31:01.039 --> 00:31:19.599
Um again, it you're seeing more of those coming in the first round of contracts uh to be reviewed, which is great, but we still see it missing from the vast majority of agreements that at least I'm being asked to review today.
00:31:19.839 --> 00:31:41.759
And so making sure that you um have an understanding of where you want to have those clauses and how they will continue through the life of that evergreen agreement, um throughout the the the at least the coming years uh while you're responsible for it, right?
00:31:42.000 --> 00:31:54.640
Um the equity and fairness arguments uh as I'm negotiating, and of course, another war story, I was negotiating a business associate agreement uh about two or two days ago.
00:31:54.960 --> 00:32:08.079
And um as we were looking at the equity and fairness, it was very clear that I was arguing against an AI um tool and not an actual person on the other line, which is fine.
00:32:08.640 --> 00:32:15.279
But uh the AI had not caught itself in talking essentially on both sides of its mouth, right?
00:32:15.440 --> 00:32:19.599
One is this is market, and so we think you should agree to this.
00:32:20.240 --> 00:32:25.920
Um and the other was, well, this is what the only what the rule says, and that's all we're gonna agree to, right?
00:32:26.079 --> 00:32:29.119
And so it was toggling between both those arguments.
00:32:29.279 --> 00:32:33.759
There was a sometimes they can coexist, sometimes they can't, suffices to say.
00:32:33.839 --> 00:32:36.720
And this one, they kind of caught themselves in it.
00:32:36.880 --> 00:32:44.799
Um, it also missed a cross-reference um to an indemnification clause it tried to remove as it was in negotiating another part of the agreement.
00:32:44.960 --> 00:32:56.720
So, you know, it's important to be thinking again about uh as we do, equity and fairness of who's most equipped to be dealing with some of these issues.
00:32:57.119 --> 00:33:09.759
Maybe it's the provider and you want control, but again, as we think about indemnification, bigger companies, uh, who's at risk, thinking about what those might be.
00:33:10.000 --> 00:33:17.440
And that's also where we uh get into some uh called debates on what the reps and warranties are going to look like.
00:33:17.680 --> 00:33:37.519
What will the um uh vendor rep to uh as it's talking about what it can say about how it will not discriminate or how it will keep up with ensuring that its uh tool is not operating in a discriminatory um uh mechanism.
00:33:38.799 --> 00:33:41.599
The privacy and security issues continue.
00:33:41.759 --> 00:34:03.200
They are just uh more complicated, and the risk now is amplified because as we have more and more data and more and more data um uh being replicated through the AI tools, more and more data being utilized for other purposes to hopefully make medical advances, right?
00:34:03.279 --> 00:34:13.920
And and and care advances, but that certainly is going to add to the risk allocation um that that we see uh quite a bit.
00:34:14.480 --> 00:34:31.519
And so the last thing I'll note too is is the complexity that can come with aligning um SLAs uh with the clinical um uh workflows uh and the reimbursement rules.
00:34:32.000 --> 00:34:38.559
Uh reimbursement uh rules in this, uh as we talked about earlier, are quite tricky.
00:34:38.960 --> 00:34:47.199
And um frankly, CMS has said we're looking for this and we we will be paying attention to this.
00:34:48.000 --> 00:34:58.480
So uh I think it's incumbent on us as we're looking at these agreements to be uh anticipating uh where that could go off the rails.
00:34:58.639 --> 00:35:20.800
Um and again, that may come back to a negotiation standpoint, uh, but we also need to be advising our uh business teams that no matter what, some of this you can't shift that risk, and it is going to be born uh to the team because that human in the loop, as Sean mentioned earlier, is hugely important.
00:35:23.119 --> 00:35:32.000
Yeah, I mean, I think that that's a great, it's it's probably not um all-encompassing, but those are those are some serious things you need to be thinking about when looking at these types of contracts.
00:35:32.079 --> 00:35:42.719
And I've experienced the same thing that a lot of these companies are developed by maybe by IT guys, you know, software guys, and they don't really have a lot of experience in healthcare.
00:35:42.800 --> 00:35:46.960
And their lawyers, if they if they're even using lawyers, their lawyers don't have a lot of experience in healthcare.
00:35:47.039 --> 00:35:52.719
So they give you a simple software contract, but it's a lot more than that, and there's a lot more things that need to be considered.
00:35:52.880 --> 00:36:03.119
And the data use and indemnification, those are all really key provisions that are very different in the healthcare context than if it's some other, you know, some other AI application.
00:36:03.280 --> 00:36:04.480
So it's a great point.
00:36:04.719 --> 00:36:09.039
I want to shift gears a little bit and and ask um ask Betsy about the future.
00:36:09.280 --> 00:36:17.440
I mean, Betsy, look forward for us and what what what should we be building into deals now for where this is headed over the next several years?
00:36:18.320 --> 00:36:27.599
Well, let me get out my crystal ball and um uh talk about a few things.
00:36:27.679 --> 00:36:49.599
Um, I think we can anticipate more regulatory development going forward, um, even though we're, you know, the current administration, I think initially a lot of people probably thought that there would be more of a deregulatory atmosphere.
00:36:49.840 --> 00:36:53.760
Um that's not really the case with healthcare.
00:36:54.000 --> 00:37:22.400
Um, so I think um, you know, it we need to be mindful um of some of those developments, obviously monitoring what is happening on with AI regulatory developments, whether there will ever be a federal AI um law, um, but also monitoring you know state AI developments.
00:37:22.480 --> 00:37:29.119
Um, you can't ignore those, and the states are you know aggressively enforcing them.
00:37:29.360 --> 00:37:39.840
Um, you know, even thinking about things like is the tool you're purchasing, could it conceivably be practicing medicine?
00:37:39.920 --> 00:37:52.719
You know, we've had the first case in uh Pennsylvania where um an AI tool has uh is alleged to have um engaged in practicing medicine.
00:37:52.880 --> 00:37:58.800
Um something, you know, that's probably something we didn't think we'd have to worry about a couple years ago.
00:37:59.039 --> 00:38:10.719
Um, you know, I we're also um, you know, you need to think about antitrust scrutiny from the FTC and DOJ.
00:38:10.880 --> 00:38:14.960
You know, a lot of the issues that um Elisa ticked through earlier.
00:38:15.119 --> 00:38:21.519
Um, you know, the FTC has a healthcare task force, so you need to be mindful if you're on the vendor side.
00:38:21.760 --> 00:38:28.079
Is your tool, is your AI doing what you you're saying it's doing?
00:38:28.320 --> 00:38:39.599
Um, you know, um, or are it, you know, could it be are your um representations about it potentially deceptive, you know, untrue?
00:38:39.840 --> 00:39:03.920
Um also increased healthcare fraud, waste, and abuse enforcement at both the federal and the state level, um, you know, which is a little surprising in the current environment, but um, you know, HHS has made it very clear that they are HHS and DOJ, that they are targeting fraud, waste, and abuse.
00:39:04.079 --> 00:39:33.840
And um, you know, while some of these AI products can deliver operational efficiency, um, they may be able to do other things at scale that could, you know, land um healthcare systems and providers um on the wrong side of um you know the OIG or CMS if you're not watching how these what these tools are doing and how they're being used.
00:39:34.079 --> 00:39:37.599
Um state oversight and private equity ownership.
00:39:37.679 --> 00:39:47.599
Um, you know, Elisa had mentioned uh PE ownership before and how that changes some of the risk um calculations when you're contracting.
00:39:47.760 --> 00:39:53.920
Um, you know, but we also expect um, you know, I think that's something to consider.
00:39:54.159 --> 00:39:56.000
Corporate practice of medicine.
00:39:56.079 --> 00:40:06.239
Uh, we're seeing a lot of there's going to continue to be a lot of activity in that area, and you know, you need to be mindful of that as you're contracting.
00:40:06.480 --> 00:40:31.199
Um, you know, FDA oversight, you know, um things you need to, especially with respect to AI-enabled medical devices, um state oversight of healthcare transactions, um, you know, um, and you know, this patchwork of state privacy laws, I think that's going to continue.
00:40:31.599 --> 00:40:43.280
Um especially as and we see more states developing or passing um healthcare data specific privacy laws.
00:40:43.519 --> 00:40:49.280
Um I mentioned the state AI laws, you know, healthcare licensing laws.
00:40:49.519 --> 00:41:01.440
Um, so there is a lot of balls in the air potentially, um, you know, which is going to make contracting um, I think, um, more challenging.
00:41:01.599 --> 00:41:05.119
Um, you know, a lot more things that need to be addressed.
00:41:05.440 --> 00:41:10.239
Also stepping um back a little more.
00:41:10.400 --> 00:41:15.519
Um, I think we're going to see more AI ready contract templates.
00:41:15.760 --> 00:41:21.599
So moving from these point solutions to platform contracting.
00:41:21.760 --> 00:41:28.239
Again, it's, you know, you're not just buying a device or software, you are getting a platform.
00:41:28.639 --> 00:41:32.719
Um, and um I think Elisa mentioned this earlier.
00:41:33.039 --> 00:41:37.119
Clients need to be ready for shorter innovation cycles.
00:41:37.360 --> 00:41:46.559
So um, you know, your has typically you may have done, you know, looked at a contract every three to five years.
00:41:46.880 --> 00:41:58.000
That cycle is going to shorten significantly with the rapidity with which new tools are developing and being improved.
00:41:58.239 --> 00:42:06.800
Um, continuous compliance obligations, um, and you know, AI is here to stay.
00:42:06.960 --> 00:42:10.960
It is going to be a permanent fixture in these agreements.
00:42:11.039 --> 00:42:19.920
Um, and so we all have to learn you know how to um live with that and help our clients adjust to that.
00:42:20.800 --> 00:42:30.000
So, but uh Alisa and Sean, um, you know, anything else that you see in your crystal balls uh going forward.
00:42:31.280 --> 00:42:43.760
I mean, I'll say that I I do think that we'll ex we should expect to see some federal AI regulation, whether that be in the Trump administration doing something informally or maybe regulatory.
00:42:43.920 --> 00:42:53.119
Um, and I don't think that the Congress Congress is gonna take it up anytime soon, but I think it's potentially over the next couple of years after these midterms that are coming up this fall.
00:42:53.199 --> 00:42:56.800
Um, so I do see foresee some sort of activity there.
00:42:58.639 --> 00:43:00.559
Gosh, I wish I were that optimistic.
00:43:00.719 --> 00:43:04.320
I'm gonna take your thing and and um I I I like that.
00:43:04.400 --> 00:43:08.719
I mean, I really do hope you're right on that because I do think it will help.
00:43:08.880 --> 00:43:17.440
Uh, you know, as much as a lot of us don't necessarily love over-regulation, some of that clarity I think would would go a long way.
00:43:17.679 --> 00:43:18.000
Yes.
00:43:18.239 --> 00:43:26.559
But um, you know, Sean, I mean, as we wrap up, do you have any key takeaways you think we should make sure we can circle back on?
00:43:27.119 --> 00:43:30.559
Yeah, I mean, I'll just wrap it up, I guess, with kind of how I started.
00:43:30.719 --> 00:43:40.559
And that is that these AI and RPM, RTM, these these contracts, these deals, these digital health solutions are not just IT contracts.
00:43:40.639 --> 00:43:47.199
There's more that needs to be thought of when diligencing these solutions and contracting with them and then ultimately using them.
00:43:47.280 --> 00:43:50.000
Um, but it's not it's not all doom and gloom.
00:43:50.159 --> 00:43:59.199
I mean, these there's just some additional considerations that we really need to think about and really think about these IT solutions with that healthcare overlay that you were talking about.
00:43:59.440 --> 00:44:18.639
And you know, think about these risks, think about the regulatory impact, think about the operational impact, think about the reimbursement impact, and and it just should be evaluated and assessed, and then ultimately that risk allocated between the parties and maybe considered in the contracting, including in the pricing.
00:44:18.800 --> 00:44:31.920
You know, you mentioned earlier, Lisa, that there that granting the use of de-identified data is something that is typically requested and oftentimes demanded by an AI vendor, but that has real value.
00:44:32.079 --> 00:44:35.679
So, how are we going to build that into the pricing for this deal?
00:44:35.840 --> 00:44:46.880
Um, I think all of those are really important considerations, but just as a takeaway, look, these are these are these are great opportunities and it's a it's a great world that we're living in to be able to access this great technology.
00:44:47.039 --> 00:44:50.960
But these are a lot more than just IT contracts, and there's a lot more to be thinking about.
00:44:51.599 --> 00:44:51.840
Great.
00:44:52.000 --> 00:44:53.039
Thank you, everyone.
00:44:53.360 --> 00:45:07.679
Yes, and I would just say, yeah, please feel free to make comments to the proposed rule that Sean mentioned, um, address on the issue of RPM and reimbursement.
00:45:07.760 --> 00:45:15.519
Um, you know, that that is another takeaway for the audience if they um are inclined to submit comments on that.
00:45:15.679 --> 00:45:18.960
So I feel we would be remiss if we did not mention that.
00:45:19.280 --> 00:45:20.079
Well said.
00:45:20.639 --> 00:45:22.559
Yeah, well, thank you for joining us for this podcast.
00:45:22.719 --> 00:45:24.239
Appreciate it, Betsy and Elisa.
00:45:29.039 --> 00:45:34.960
If you enjoyed this episode, be sure to subscribe to AHLA Speaking of Health Law wherever you get your podcasts.
00:45:35.119 --> 00:45:51.840
For more information about AHLA and the educational resources available to the health law community, visit American Health Law.org and stay updated on breaking healthcare industry news from the major media outlets with AHLA's Health Law Daily Podcast, exclusively for AHLA comprehensive members.
00:45:52.079 --> 00:45:59.039
To subscribe and add this private podcast feed to your podcast at once, go to americanhealthlaw.org slash daily podcast.com.
