Artificial Intelligence (AI): How To Use It And What To Beware Of With Sharif Zeid Of Empower EMR

[00:00:00] There's a cost aspect to all of this as there is with anything and just trying to figure out really where the value is actually going to be for people.

[00:00:07] I mean, it's great to read about it in the newspaper and this is going to change the world.

[00:00:10] But at the end of the day, we've got practices, we've got considerations of staff and we've got considerations of budget and we've got considerations around what actually can be implemented in a practice and in what timeframe and so forth.

[00:00:25] Welcome. You've entered the Physical Therapy Owners Club podcast, where your host, Nathan,

[00:00:30] Shields and other successful PT owners and leaders share their experience and insights on how to build successful PT businesses.

[00:00:37] They'll share the stories of their paths to success and show you how you can also obtain greater freedom and more profits from your business.

[00:00:44] That's what the PT Owners Club is all about, greater freedom and more profits.

[00:00:49] There's plenty of room for you as well, so come on in and join the club.

[00:00:53] This episode is brought to you by Empower EMR, a modern AI driven EMR that not only keeps owners in control, it keeps billers happy, PTs, OTs and speech therapists productive without burnout and keeps your patients happy as well.

[00:01:16] Empower EMR lets you automate workflows so you can focus on delivering quality care profitably.

[00:01:23] We are happy to have Empower EMR as a sponsor of the Private Practice Owners Club and you can learn more at Empower EMR dot com.

[00:01:32] Hello and welcome to the Private Practice Owners Club podcast.

[00:01:35] I'm Nathan Shields and got one of our sponsors for the year, Empower EMR and Sharif Z, the founder and CEO of Empower EMR joining us.

[00:01:45] Thanks for joining us today, Sharif.

[00:01:47] Hey, nice to be here.

[00:01:47] Thanks for having me back.

[00:01:48] Appreciate it.

[00:01:49] Yeah, you've been on the podcast a couple of times already.

[00:01:52] And thank you so much again for being a sponsor of our, our first ever annual conference back in September and Clearwater is great to have you there and, and have your presence and have your support.

[00:02:02] No, totally my pleasure.

[00:02:04] And just as a plug, if you know, I know there, there's more to come on events.

[00:02:07] So if you're listening to this and you're, you didn't come to Clearwater and you're contemplating what to do for 2025, whatever Nathan and the team put together should be on your list, wherever it is, even if it's in the frozen tundra of Alaska or something like that, you should go.

[00:02:20] So it's that good.

[00:02:21] Yeah, we wouldn't do that to people, but unless it was in the summer, that would be an appropriate place, but it's still a hall.

[00:02:26] So thank you for the plug.

[00:02:27] I really appreciate that.

[00:02:29] And that's very thoughtful, but it's good to have you on and it's good to catch up with you a little bit.

[00:02:33] And specifically from your perspective, what we're talking about today is artificial intelligence, AI, and its influence on our industry, on our business ownership, in general, how it can affect us both positively and what to look out for negatively.

[00:02:49] And so you coming from the tech side of things and having that tech background over the past number of decades, you've been in this physical therapy space specifically for a couple of decades already.

[00:03:00] So you can see where it's coming from, where it's going, what to look out for, what to be mindful of and what to actually look forward to.

[00:03:09] I'm looking forward to this conversation just to say it that way because of your perspective.

[00:03:13] And so where would you like to start with this conversation?

[00:03:15] Do you want to start like, where did AI come from and move forward chronologically?

[00:03:19] Where do you want to start?

[00:03:20] Yeah, I think that could be a great place to start.

[00:03:22] I think, I mean, I guess to give you my perspective, even from the tech side, it seemed like you went to bed one night and AI didn't exist.

[00:03:30] And then you woke up the next morning at like 8 a.m. and suddenly everything was AI, like everything.

[00:03:35] You're eating your...

[00:03:35] In blue.

[00:03:36] Yeah, your Eggo waffles had an AI logo on the front of it.

[00:03:39] And you're like, how does this make any sense?

[00:03:40] These are waffles.

[00:03:41] So, you know, it's interesting.

[00:03:43] And this is very recent, of course, in the last couple of years.

[00:03:45] If you, you know, you read any sort of news that any major publication newspaper, I mean, again, it just took off like wildfire.

[00:03:52] And really a lot of that came from, you know, AI isn't exactly new, but, you know, the sort of presentation and certainly the applications of it and the accessibility of it today are things that have changed dramatically in the last couple of years.

[00:04:04] And I think the seminal moment in sort of in this story was really OpenAI, which is a company releasing ChatGPT.

[00:04:11] And then it just sort of all of a sudden it was on everybody's computer.

[00:04:15] Previously, you know, you would see it very sort of locked up and maybe in a government type application, a military type application or for very large corporations.

[00:04:24] And another part of this sort of whole thing is that they were often previously, they were often trained on smaller sets of data compared to what they're what's considered a large set of data today.

[00:04:33] And that's because whoever was running it was training it.

[00:04:35] Now you're basically dumping in all kinds of stuff into it from all corners of the web and all corners of now people using it.

[00:04:42] And so it just really got expansive.

[00:04:44] And it's a natural evolution of you, if you will, of the last cool buzz tech word, which was big data.

[00:04:50] You know, if we were doing this podcast five years ago, everybody was like, we're doing big data.

[00:04:54] And I always felt back then that many people knew the term big data.

[00:04:58] Nobody was really knew what they were doing with it or how to use it.

[00:05:01] And many didn't actually have enough data to be qualified as big data.

[00:05:05] You know, Walmart has big data.

[00:05:07] Google has big data.

[00:05:08] Do you have big data in your practice?

[00:05:09] Probably not.

[00:05:10] But you now have access to it in at least in an indirect way through some of these AIs.

[00:05:15] And it's really cool and a little bit scary.

[00:05:18] And certainly there are ethical questions and logical questions and business questions.

[00:05:22] And just trying to, that's what I hope we get into a little bit.

[00:05:25] It's just trying to like make sense of all this and take it from this huge headline you're reading in the newspaper to like, how could I use this?

[00:05:32] Maybe where should I not, you know, maybe where should I pump the brakes a little bit and either wait for it to evolve or simply avoid it for certain purposes.

[00:05:38] And it's incredible.

[00:05:40] And things move so fast now.

[00:05:41] So what can you do?

[00:05:42] Right.

[00:05:43] And maybe that's a good place to start is if you had the chance, if you had a room full of private practice owners or small business owners, frankly, where would you recommend they start?

[00:05:53] If people are saying, you know, AI is the future.

[00:05:55] You need to learn how to use it.

[00:05:57] For those people who maybe want to dip their toes in that pool, where would you recommend they start with it?

[00:06:03] Well, here's what I would say.

[00:06:04] I mean, I think the first thing with anything new, like let's say a new clinical technology came out, you know, forget about AIs.

[00:06:10] Let's talk about what PTs really know, right?

[00:06:13] Clinical side of things.

[00:06:13] If a new machine comes out, a new technique comes out.

[00:06:17] I think the first thing is always some sort of measure of caution, right?

[00:06:20] We don't want to do any harm.

[00:06:21] We don't certainly don't want to hurt a patient or anything crazy like that.

[00:06:24] We don't want to hurt our business.

[00:06:25] And so we want to be, you know, naturally skeptical about things that are out there.

[00:06:30] And, you know, part of AI is really that it needs data, tons of data to be useful.

[00:06:35] And one thing we certainly don't want to do is either inadvertently or purposefully expose data to an AI that we don't want it.

[00:06:42] Say, for example, a patient's information.

[00:06:44] That's like, you know, that's a no.

[00:06:45] We don't want to do that unless we have some confidence that, you know, okay, this thing is really meant to be taking this.

[00:06:52] And that it's not going to then spit that out to somebody else's answer two hours later and say, oh, yeah, Nathan has the following thing going on with him.

[00:06:57] And it because it just learned that it's AI, it's smart, but it's not.

[00:07:01] But it's dumb right at the same time in this sort of so it doesn't have discretion necessarily.

[00:07:05] Now, this is also rapidly evolving.

[00:07:07] There are AIs now that are meant to can train on large sets of data and then you can have some measure of privacy.

[00:07:14] And then, of course, many companies like ours and others are building around AI, you know, foundational elements of AI technology and then presenting you with just something that's ready to use.

[00:07:25] And we've kind of helped handle some of those things.

[00:07:27] So that's where I would go with where do you start with this is, of course, commercially available, you know, AI tools.

[00:07:33] Right now, I think scribes are one of the biggest area of interest and intrigue.

[00:07:37] I mean, how many podcasts have there been on your episodes where you're talking about documentation hassles and burnout and all these sorts of things?

[00:07:44] And scribes are really, in my humble opinion, one of the most meaningful potential movements in the last even 10 years on really moving document, moving the needle on documentation from a timeline.

[00:07:55] And there are lots of great opportunities to, you know, you're not going to build your own scribe.

[00:08:00] Listeners are PTs, right?

[00:08:01] They're not technologists.

[00:08:02] They're going to build it.

[00:08:03] But there are lots of commercially available options and rapidly expanding in their abilities and capabilities.

[00:08:09] And that's like a great place to consider starting.

[00:08:12] So that's like, let's put that on the clinical side.

[00:08:15] Another area would be forget about clinical.

[00:08:17] There's all these other functions in your business that can benefit from AI and don't have to deal with the HIPAA aspect or the compliance aspect.

[00:08:25] For example, let AI write a job description for you.

[00:08:28] Maybe give it a few things.

[00:08:29] Let it spruce it up for you.

[00:08:30] Or maybe let it write it from scratch.

[00:08:32] Maybe help with your website content.

[00:08:35] You know, have maybe produce a blog article with it or clean up the copy on your homepage of your website and things like that.

[00:08:42] So I think there are some really nice, modest ways that aren't like a six-month commitment, a $6 million commitment that you can just jump in and do it for your jobs, do it for your website, try a scribe out, get some of your PTs going.

[00:08:55] And then really start to see some actual meaningful changes in your day-to-day.

[00:08:59] Yeah.

[00:09:00] No, I think a lot of people, when we initially considered how AI might be implemented into the world in general, we immediately thought that it was going to automate industrial processes, right?

[00:09:13] Robots taking over low-level blue-collar jobs, right?

[00:09:17] And I think some of the people in the world are kind of surprised that it's doing the opposite.

[00:09:22] It's taking over some of the creative roles in our society, content creation, art, music, et cetera.

[00:09:30] AI can produce these, what used to be creative projects and do very well at it.

[00:09:36] And I thought maybe that's where you're going to go initially.

[00:09:39] I love that you brought up the scribes, but if someone were to ask me, where should I start?

[00:09:43] I would say, you know what?

[00:09:45] Start creating marketing content with the help of a chat, GPT, Claude, AI, whatever is out there.

[00:09:52] I personally actually like Claude better than chat, GPT for some of that stuff, but that, you know, to each their own.

[00:09:58] It doesn't really matter, but you can do either the free version or you can do the $20 a month version.

[00:10:04] And the $20 a month version allows you to like upload some of your content and ask for it to ask for recommendations and stuff like that, and even maybe create some design work.

[00:10:15] So there are some great opportunities with AI that can actually take a lot of brain energy and time off of your plate by simply creating blog posts, creating content for a newsletter, if you will.

[00:10:28] Like you said, upload some of the content from your website and say, how would you recommend changing this in order to improve my SEO?

[00:10:36] You know, and guide it and direct it for the things that you're wanting to get from that content.

[00:10:41] I know there are tricks and certain prompts that will help it create things better for you.

[00:10:47] And that might take a little bit of training, but it's worth doing a little bit of Google research, frankly, to say, hey, how do I properly prompt my AI system?

[00:10:58] Give me what I want.

[00:10:59] What are some of the prompts that I need to know and leverage to get the best responses?

[00:11:03] And so that's what's really helped me is, you know what, I need to do a presentation on blank.

[00:11:09] Give me the top four or five things that I should cover that people are worried about in this industry and give me a 15 minute presentation.

[00:11:18] And bam, you've got at least the outline right there to work off of.

[00:11:22] And it does a really, really good job.

[00:11:25] Now you're going to have to customize it a little bit and that's fine.

[00:11:28] So it doesn't sound like robot speak, but it is a place to start.

[00:11:32] But I also appreciated that you said specifically to our industry using the AI businesses that are out there.

[00:11:39] I've had prediction health.

[00:11:40] I've had comprehend health both on the podcast in the past.

[00:11:44] And they're worth looking into to improve your documentation because we all know the number one headache in the industry is documentation amongst the providers, right?

[00:11:55] Providers don't want to go home with documentation.

[00:11:57] They don't want me doing it on the weekends.

[00:11:58] They would love to walk out the door at the same time as their patients with all of the documentation complete.

[00:12:04] And these dictation scribe services that leverage AI are a great way to do that.

[00:12:10] They are worth the money to make sure that your notes are done in a satisfactory way.

[00:12:16] They'll also review for compliance.

[00:12:18] They'll also make recommended CPT code changes to your benefit.

[00:12:23] Frankly, if their activity is better than a therapeutic exercise based on your documentation, it can recommend that and the reimbursement is greater.

[00:12:31] So those kind of things are ways that we can leverage AI as owners to benefit us and make things easier on us and our teams.

[00:12:40] And I love that you brought up both of those aspects.

[00:12:44] Yeah, it's amazing.

[00:12:45] And there's a couple of things I wanted to key in on there.

[00:12:47] One was you bring up a good point.

[00:12:49] There are various AIs out there, if you will, if you're going to go with a very like a straight generative AI, as it's called.

[00:12:55] And generative AI is the part of AI that's really generates, right?

[00:12:59] Generates like what you said, text, copy images, things along those lines.

[00:13:02] And that is what really everybody's talking about these days with AI.

[00:13:06] There are other kinds of AI and they're not as applicable to as wide an audience, which is why you hear about them.

[00:13:11] And so if you go to some of these, you know, the ChatGPT, Claude, you have Grok, you have all these different ones.

[00:13:17] What I would say in terms of preferences, yeah, you learn which one you like.

[00:13:22] And they're really what's underneath that is called an LLM, which is a large language model.

[00:13:27] And that's what's basically how it learned.

[00:13:29] And so there's a reason why you might like one better than the others, because it's just like it learned a little.

[00:13:32] It went to a different school and it got a different, slightly different degree.

[00:13:35] And it just, it kind of, you jive with it and you like it and you kind of feel like you're hitting it off with them.

[00:13:40] So you made a new friend.

[00:13:41] And to your other point, the other one I wanted to mention was, yeah, I mean, when you use one of these straight up generative AIs where you just, it just gives you a box and you just type in whatever you want to enter and then it gives you back something.

[00:13:51] Yeah.

[00:13:51] I mean, what you want to do is think about it like a, like a sort of a smart person, but they have to be told like as specifically as you can, what you want them to do within reason.

[00:13:59] So if you ask it to write something, you can try interesting things like make it upbeat or use a tone that's serious or, you know, whatever you want to do.

[00:14:07] And you can try it a couple of different ways and you find your voice, if you will.

[00:14:11] So your website might have a mantra, right?

[00:14:14] It's like an upbeat website or it's more of a business, you know, the verbiage is more just straightforward and formal and literal.

[00:14:20] And you find whatever you want.

[00:14:22] People use it to write emails too.

[00:14:23] And so they, they, they find the same kind of concept with it.

[00:14:26] So I love the idea of marketing stuff.

[00:14:28] I mean, I come across, unfortunately, a lot of sad websites that need an update, whether it's, you know, a big update or a little update like hasn't been touched in years.

[00:14:37] And, you know, it's just a great shortcut is how I look at it.

[00:14:40] It's not going to cut out the work and do it a hundred percent because it's unlikely what you get out of it will be immediately ready to just post.

[00:14:46] But it could, the better you get at it, it could be very close.

[00:14:48] A little bit of editing, a little bit of cleanup, make sure it's saying what you want it to say.

[00:14:52] Then you're just copying and pasting.

[00:14:54] And so the excuses sort of for not update, keeping your website up to date or just putting something fresh on there every once in a while, I think diminish a little bit.

[00:15:02] And again, no HIPAA there, no stuff.

[00:15:04] So it's like, it's easy to go.

[00:15:06] And I do think paying for some of them is worth it if you're really going to get into it.

[00:15:09] But I think trying the free one, free option is just a great way to dip your toes in and see if you, you can find a use for it in your day-to-day week-to-week business life.

[00:15:17] Yeah.

[00:15:18] You know, one thing that was really helpful for me as I was promoting something and I, and I needed say 10 or 12 posts for social media that I wanted to put out regarding something that I was selling.

[00:15:29] And I said, here's what I'm selling.

[00:15:31] These are some of the benefits of it.

[00:15:33] This is, I want it to be a focus on blank, blank, and blank.

[00:15:38] And this is, and I need 10 posts and boom, boom, boom.

[00:15:42] Shout out 10 of them.

[00:15:43] I liked maybe four or five of them.

[00:15:45] And I was able to say, you know, I like these four or five create four or five more of these types.

[00:15:51] Right.

[00:15:51] And it was okay.

[00:15:52] Bam, bam, bam, bam, bam.

[00:15:53] At the time it would have taken me to write out those 10 posts comparatively, I saved so many hours.

[00:15:59] Oh my gosh.

[00:16:00] And I always say too, I mean, if you remember back to harken back to your days in like maybe college or high school, you had to write a paper.

[00:16:06] The art of the start is often the hardest part.

[00:16:08] So even if it just gives you like a spark, that is a huge thing.

[00:16:12] You know, if it's even, it's imperfect, but the idea is not bad or this first sentence is not bad.

[00:16:16] It's like, gosh, it's so much easier to write five more lines once the first line has been written than it is to look at a blank word document and say, okay, I'm going to write this thing from the ground up.

[00:16:25] So, so it's tremendous.

[00:16:26] And you can do this today.

[00:16:28] I mean, you can jump on whatever, try it out and, you know, see how you do.

[00:16:32] And by the way, what you're talking about with, you know, how you prompt the AI is it's actually becoming a job.

[00:16:37] It's called the prompt engineer.

[00:16:39] And that's like an actual position that people are now studying in college.

[00:16:42] Now these folks are generally probably more on the other side of like actually building how the system responds to prompts, but also they do.

[00:16:49] They create inputs too.

[00:16:51] So it's that it can get that complex.

[00:16:53] It's not going to be for any, I think for this application, it's just simple.

[00:16:56] Keep it simple.

[00:16:57] A few little tips though, on like just telling it what tone you want, or if you want it to be long or short, we'll give, we'll get you even closer to that kind of ready camera, ready, ready to go, ready to be published.

[00:17:06] Well, we can take it even a step further for those people who are looking for providers.

[00:17:11] And I know a majority of people are in the current environment as they're looking to recruit.

[00:17:17] Why not throw it out there?

[00:17:18] Like, give me a four-step recruiting process that will provide me an ideal candidate with these values.

[00:17:26] And these are some of the benefits I provide for my opening.

[00:17:29] Give me an appropriate marketing strategy to find these, this person.

[00:17:33] And it's going to actually, it's going to give you some ideas and probably half of them are things that you might not have thought of.

[00:17:39] Absolutely. And remember, it's, it's read, you know, 10 billion resumes across or, or job postings across every industry.

[00:17:46] So, I mean, you, you get a flavor of thing.

[00:17:48] If you're just like, okay, let me go look at my competitor across town or across state.

[00:17:52] What did they write in their job post?

[00:17:54] You just, your world is so much bigger and that's just, that's a tremendous benefit of it for sure.

[00:17:59] I mean, you can do that for recruiting.

[00:18:00] You can even say, build out a marketing strategy for me for the next year.

[00:18:04] Like, what do I need to do to get more patients in the door?

[00:18:09] What are other providers doing?

[00:18:11] And what are some of the avenues that they're using to find new patients?

[00:18:15] And that, I think those are just ways to like use it to, so you don't have to come up with your own stuff.

[00:18:21] I mean, I've got stuff out there, but you have to find my stuff.

[00:18:24] Whereas you can just punch it into chat GPT, then it'll bring it to you.

[00:18:28] 100%.

[00:18:28] Yeah.

[00:18:29] No, please.

[00:18:30] If you're listening, I wouldn't say stop listening, but go do it.

[00:18:32] Just try it.

[00:18:33] You got, you got nothing to lose really.

[00:18:34] If you don't end up loving it or it doesn't produce anything for you.

[00:18:37] Fine.

[00:18:38] But you know, you're literally talking about 10 minutes.

[00:18:40] Just jump on and ask any question or better yet.

[00:18:42] If you have kids like I do, they have an endless number of questions that we seem to ask,

[00:18:47] you know, how big is the ocean or why is it this way?

[00:18:49] Or why is this, this thing?

[00:18:50] And chat GPT is interesting at giving answers as well as other tools as well.

[00:18:56] Well, tell me a little bit about how AI is influencing the EMR space from where?

[00:19:01] Yeah.

[00:19:02] Well, I think there's not a corner of the space that isn't being touched at least tangentially by the concept of AI.

[00:19:09] Obviously, tremendous amount of interest from, you know, clients and from those who are maybe considering our product for their practice.

[00:19:15] But I do think there's some learning curve happening in terms of understanding what exactly is this thing in terms of practical applications for my practice.

[00:19:23] What does it mean?

[00:19:24] I also think that there's a cost aspect to all of this as there is with anything and just trying to figure out really where the value is actually going to be for people.

[00:19:32] I mean, it's great to read about it in the newspaper and this is going to change the world.

[00:19:36] But at the end of the day, we've got practices, we've got considerations of staff and we've got considerations of budget and we've got considerations around what actually can be implemented in a practice and in what time frame and so forth.

[00:19:48] So I think right now, if I were to characterize it very much like it's not quite like the days I was here when EMR became a thing.

[00:19:54] I converted hundreds of practices, thousands perhaps from paper to electronic.

[00:19:59] That was such a wild, wild west notion.

[00:20:01] You know, A, practices were taking on an expense they never had had before.

[00:20:06] You know, like, well, I'm replacing paper and pencil with a bill.

[00:20:08] And why should I do that?

[00:20:09] But there are some elements that I observed back then from early adopters to, you know, the curve of adoption where some people are way early and we're using CHAP GPT the day it came out.

[00:20:19] Then you kind of got the middle where a lot of people will start to just adopt because they become more standardized and then you've got the late adopters who are like really just hanging on.

[00:20:27] It's like finding a unicorn practice today that still writes its notes on paper.

[00:20:31] They do exist.

[00:20:32] There's not a lot of them.

[00:20:32] That's why we call them unicorns.

[00:20:34] But every once in a blue moon, we'll hear about that.

[00:20:36] So I think we're very much still in the early part of that curve.

[00:20:38] And I think for the industry itself, you know, on the tech side, it's really trying to find out where can we make meaningful advances for practices and how can we make those in a cost effective manner.

[00:20:50] And I'm finding that as usual, you know, the market's going to do what it wants.

[00:20:53] I think there are some gimmicky things personally that I don't think are that meaningful and they just get splashed out all over the place.

[00:20:59] And then I think there are some real true value ads that are being done out there.

[00:21:04] And you mentioned a couple of them.

[00:21:05] We talked a little bit about the scribe.

[00:21:06] I think the compliance is a tremendous area where you're talking about the traditional chart review model.

[00:21:13] You know, Nathan pulls 10 charts a week or whatever.

[00:21:16] Or worst case, Nathan never pulls any charts and never does any compliance.

[00:21:19] And so you can replace that with a, hey, can an AI give guidance?

[00:21:23] And that's already evolved in the last year where now we're backing into not just retrospective, looking back at notes that have been signed, but also in real time, providing feedback about, you know, are you doing a good job with your note?

[00:21:34] Does your coding look correct?

[00:21:37] And so I think it's just kind of, it's not the wild, wild west, but maybe it's like the gold rush now.

[00:21:42] There's the wild, wild west.

[00:21:43] That was the opening of the frontier.

[00:21:44] Then you've got everybody saying, okay, I think there's some value here.

[00:21:47] We've got to find where the value is.

[00:21:48] And that's what our commitment is really is to just saying, okay, we've been in this space for a long time.

[00:21:53] We really understand the plight, I think, of the PT practice.

[00:21:56] They're getting squeezed from a lot of different places.

[00:21:58] You know, reimbursements are a constant conversation.

[00:22:00] But then there's the idea of, okay, we've got to make headway on this burnout problem with PTs.

[00:22:05] We've got to make headway on our compliance.

[00:22:07] And so the last thing I'll say is just broadening it out from just EMR is that all of healthcare is dabbling in AI in some form, right?

[00:22:14] And that means insurance companies too are implementing their own stuff.

[00:22:18] The downside, I think, for practices is they have more data because they get to amalgamate everybody's data.

[00:22:24] And so I think it'll be interesting to see how that plays out over time.

[00:22:28] And if it, you know, if it's AI is kind of bumping into each other, I think it'll be interesting to see how industry, large industry groups can maybe help be a counterbalance to some of that counterweight by helping aggregation and things like that.

[00:22:39] So that there can be a response and an equal balancing of that relationship, which is forever in a day been a challenge, but just in different ways.

[00:22:48] Now it's a technology potential imbalance.

[00:22:51] So it's a really, it's a time of really, it's a great time.

[00:22:54] It's an interesting time.

[00:22:55] It's an active time.

[00:22:56] But it's also a time like I sort of opened with saying, all right, you know, like what does make a meaningful difference?

[00:23:02] Just going out and spending extra thousands of dollars a month or a year or whatever.

[00:23:06] If it doesn't provide any meaningful change for your practice, probably isn't the right thing to do.

[00:23:11] Yeah.

[00:23:11] And I think like you mentioned earlier and to reiterate, starting from a scribe perspective is a fairly simple way to get into it clinically.

[00:23:20] I, back in the day, I used to use Dragon and get transcription that way.

[00:23:24] And, you know, voice to text kind of stuff.

[00:23:27] And that was helpful.

[00:23:28] It had its issues, but it's so much better now.

[00:23:30] And the fact that AI learning language model can learn your language as you're doing the documentation and then write how you tend to speak and kind of learn that and get better over time can make it very customized to you or any other provider specifically.

[00:23:48] It's a great place to start.

[00:23:49] And when you consider that the insurance companies are going to be using AI to their advantage, we better be getting used to it in order to combat whatever they come up with.

[00:24:01] Because we're going to lean on our EMRs and your amalgamation of data as well to say, you know, this is what companies, the insurance companies are doing.

[00:24:10] This is what we need to do to respond to that and make that an AI response across the board.

[00:24:17] Right.

[00:24:17] Yeah, absolutely.

[00:24:19] And you kind of keyed in on an interesting topic, which is, I think, true for all technology, but especially for AI is there's often an expectation that or desire that, you know, if it's not perfect, then I shouldn't implement it.

[00:24:30] I should wait on that adoption curve until it's perfect.

[00:24:33] But I'm always trying to evangelize this position to folks, which is incremental gain is so important.

[00:24:39] You know, if let's talk about incremental gain in a couple different ways.

[00:24:42] One could be that you have 10 PTs in your practice and six of them really take to this new technology and four don't.

[00:24:50] You should still move forward if you can, because 60 percent of your staff say having less burnout is a huge win.

[00:24:57] Would we like it to be 100 percent?

[00:24:58] Absolutely.

[00:24:59] We'd love it to be 110 percent.

[00:25:01] But 60 percent is pretty darn good.

[00:25:03] You know what?

[00:25:03] In some cases, so is 20 percent.

[00:25:05] You know, if you've got five PTs and you can't really afford to lose one, if that person says, you know, this technology will help me and I will use it.

[00:25:13] And you're like, well, I don't know if I want to write the check for it.

[00:25:16] Figuratively, I don't know if I want to pay for it.

[00:25:17] You probably should, because first of all, the cost relative to losing, say, losing that employee to burnout or losing them is tremendous replacement cost.

[00:25:27] And so I hear this all the time.

[00:25:28] You know, we have some there.

[00:25:30] So that's one example is where you have a split in your clinical staff.

[00:25:33] Some are like tech savvy and are good with stuff and others can't really change as hard for them.

[00:25:37] Fair enough.

[00:25:38] Another thing is, let's say you use an AI to help you write your note and, you know, or you use an AI to write your marketing thing and you've already kind of hit.

[00:25:45] It's like 80 percent good.

[00:25:46] Well, you just saved 80 percent.

[00:25:48] So, again, I would love it to go all the way and go from the clod right into my website.

[00:25:53] But, you know, we've got to have a little bit of leeway here.

[00:25:55] And so but these are still both huge improvements.

[00:25:57] And we do see a struggle with that because sometimes the decision making gets hung up on sort of the least the lowest adopting person's sort of perspective on it as opposed to the highest adopting person's.

[00:26:09] And we really recommend, you know, wherever you can.

[00:26:12] And this is, again, not just for AI, but anything with your EMR, like push a little bit and get those people going and then try and get those in the middle to kind of move up.

[00:26:20] And then eventually the late adopters will kind of fall into place on their own in due time.

[00:26:24] But take the win if you can, especially if it's substantive.

[00:26:27] Yeah, there's something to be said for even if there's one out of your five providers that was leveraging, say, a scribe and the AI scribe specifically.

[00:26:38] And they were able to get their notes done faster.

[00:26:41] And they are and you know that they're compliant.

[00:26:43] And so they are walking out the door when their last patient walks out their door, just turning off the lights as they go.

[00:26:50] And while the other providers see that, maybe they don't jump on initially, but they see that over a day or two or a week or a month.

[00:26:59] And they're like, wait, I'm still doing my stuff while you're walking out the door with your patient.

[00:27:03] That's just not right.

[00:27:05] And that's when, OK, maybe I need to take a dip in that pool as well.

[00:27:09] So just having one out of the five is enough to make an influence.

[00:27:14] Seeing is believing for sure.

[00:27:15] And I use the example of the paper to EMR switch from years ago.

[00:27:19] I mean, we did.

[00:27:20] We had practices that five PTs in the practice, three could get on board with documenting electronically, which now just seems like a hilarious thing to talk about.

[00:27:28] And two couldn't.

[00:27:29] And, you know, we're like, just got to keep going.

[00:27:31] If you're going to wait for the two, you may be still on paper in 2024.

[00:27:35] So, yeah, it's got to move forward.

[00:27:37] And I think a lot of these things are pretty approachable, but some of them do require, you know, a little bit of a change of thinking, I guess.

[00:27:44] You know, I mean, you've got to kind of wrap your head around this AI concept and that and that's part of where playing with it can show you that it can do stuff.

[00:27:51] I mean, I'm a technologist, but I'm also naturally a skeptical person.

[00:27:55] So you have to show it to me.

[00:27:56] And so you go and you try it and then you try it a couple more times and you say, you know, God, this is actually pretty good.

[00:28:02] There is some merit and some value here and there's some limitations as well.

[00:28:05] But, yeah, seeing and trying is believing, as they say.

[00:28:08] Yeah.

[00:28:08] Where do you see some AI affecting us going forward, whether clinically or personally?

[00:28:14] Anything in the future that you can foretell?

[00:28:16] Well, I think what's interesting right now is I think the core technology is what you have today is probably going to be defining for some years to come, meaning like the concept of a large language model, the concept of having a generative AI is here.

[00:28:30] And that's not, you know, those fundamentals aren't going to change.

[00:28:33] I don't think they're going to expand.

[00:28:34] My take is I don't think they're going to expand exponentially, say, in the next year or even two years.

[00:28:39] What I do think is going to grow and blow up and it's already happening is just people finding tons of new applications for this technology.

[00:28:46] How can you use it?

[00:28:47] What can you do with it?

[00:28:48] How can you skin it?

[00:28:49] I mean, and again, this is far beyond the confines of health care.

[00:28:53] But I mean, Apple releases the new iPhone, whatever it is, 16, I think, and they sort of embed AI in it.

[00:28:58] And then they call it Apple intelligence, which also is AI.

[00:29:01] I'm sure that was no irony in that.

[00:29:03] But maybe the AI came up with that.

[00:29:04] Who knows?

[00:29:05] But, you know, they're baking it in.

[00:29:07] And the idea is like it can do it.

[00:29:09] They sort of have threaded it throughout the entire iPhone experience is what they say.

[00:29:13] But if you're using it practically, it can like do stuff in email.

[00:29:15] It can do stuff on this.

[00:29:17] It can do stuff in this app.

[00:29:18] So it's not like a single place, but rather it's threaded throughout.

[00:29:21] And I think for EMRs, that's kind of the same concept you're going to see is we're going to continue to innovate and discover new applications.

[00:29:29] Documentation was a very natural first place to go.

[00:29:31] But we're already seeing, you know, on our end, innovation in the billing realm and trying to do automations and AIs that can do things that humans were doing that are kind of repetitive tasks in some cases, but need a little bit more of a brain, I guess, than just like the traditional like, all right, just do this 500 times in a row.

[00:29:48] So, yeah.

[00:29:49] Yeah.

[00:29:49] Maybe like insurance denials or if your code gets kicked back, you know, learn that that code can't be billed instead of the provider having to remember that every time.

[00:29:58] Maybe that code for that insurance just cannot be billed and that's learned and embedded in the system.

[00:30:04] Exactly.

[00:30:04] And that's where a great example of where more data helps.

[00:30:07] I mean, the more you can quote learn from, the more solid what you have learned is.

[00:30:11] And so, so I think billing is a natural place to look to because it's high volume with respect to codes and adjustments and write-offs and patient responsibility and all that stuff.

[00:30:21] But again, it's everywhere.

[00:30:22] And so I think that's the innovation you'll see.

[00:30:25] I also think you'll, you'll move away a little bit from, and this is already happening from just like the box you just type into.

[00:30:30] Again, as you get specialized applications, the developer of that has already kind of decided what its purpose is going to be.

[00:30:37] So they're going to curate what you input, you know, to get the output as opposed to just saying, here's a box, type anything you want, which people can struggle with.

[00:30:43] It's, it is hard to like, sometimes just here's a box.

[00:30:46] Like Google, everybody understands that.

[00:30:48] You go to Google, you type in the box, what you think you want to find.

[00:30:50] Yeah.

[00:30:51] You kind of do that with AI, but you can get better results by being a little bit more thoughtful about what you put into there.

[00:30:57] You know, last question I have, and this just came to mind is, do you foresee a point where AI might review your plan of care or your interventions and make recommendations based on current studies or anything like that?

[00:31:13] That might say, you might want to consider implement adding this to this patient, considering their diagnosis and their presentation and what you're currently doing.

[00:31:21] Or do you see maybe that possibility coming into place or is that not a function of an EMR?

[00:31:28] Yeah, really, really, really interesting question, right?

[00:31:31] So now I think you really are into medical legal questions.

[00:31:35] You're into ethical questions.

[00:31:36] So let me answer your question directly first.

[00:31:38] Yes.

[00:31:39] I mean, I think from a tech perspective, there are already efforts out there to try and say, diagnose people with AI, right?

[00:31:45] You put in, you have patterns of disease and then you say, okay, here's what patients present with objectively or subjectively.

[00:31:51] You have a million cases.

[00:31:52] Then it's like the AI becomes like the dog that can figure out you have cancer.

[00:31:56] It just, because of what it's learned.

[00:31:57] Yeah.

[00:31:58] Just to, I don't want to get tangential, but could you put in some of those things into chat GPT now?

[00:32:04] Will it give you a recommended plan of care or will it not go that far?

[00:32:07] You know, I can't say I've tried it.

[00:32:09] So I think that's, we're going to assign that as homework, but do not use real patient just as the classic disclaimer.

[00:32:15] Don't copy and paste one of your notes into chat GPT.

[00:32:18] Let's make a fake patient.

[00:32:19] Yeah.

[00:32:19] But somebody could give that a try.

[00:32:21] So again, from a tech perspective, I mean, what could AI possibly be better at than reading thousands of pages of studies and potentially thousands of pages of notes and trying to coalesce that in its learning model and to say, okay, these things go together.

[00:32:34] And so these are patterns and then you plug it in and it says, Hey, you might want to try this modality or this exercise could help.

[00:32:40] But this is actually an interesting question more for PTs who have spent seven years or more or less getting a DPT.

[00:32:47] And then it's also a question for patients.

[00:32:49] Do you know, where do you put your trust into your provider, right?

[00:32:52] If you go to your doctor and they say, well, you know, I checked with chat GPT and it says, I should probably give you the shot.

[00:32:57] You're probably not feeling terribly good about it.

[00:32:59] So this is an area where we have a lot of questions philosophically about where this is all going to go.

[00:33:04] But I personally do think that I think a hybrid model here where you have the provider, obviously, who's trained and experienced and is a human holding another human's life in their hands, so to say, or their care or their health in their hands combined with the benefits and power of really just large amounts of data could be helpful.

[00:33:22] Yeah.

[00:33:23] We'll see it now.

[00:33:24] The last interesting thing on that is always liability.

[00:33:26] But here's an interesting analog for you to it, which is you kind of see this already, but it's just happening as it goes, which is what about self-driving cars?

[00:33:35] You know, if the self-driving car crashes into another car, who owns that problem?

[00:33:38] And so in some ways, the industry has been pushing that faster than the law can catch up in some cases.

[00:33:45] I don't know.

[00:33:45] I think out your way where you are now, there's a lot of Waymo.

[00:33:48] I'm out on the East Coast.

[00:33:49] We don't really see Waymos around here.

[00:33:51] A ton of Waymos.

[00:33:52] Nobody's in the car, but it's got all these crazy contraptions on it.

[00:33:56] It really looks like what you thought sci-fi was from 30 years ago.

[00:33:59] But then Tesla introduces their new taxi thing, whatever that's called, and it doesn't have all the...

[00:34:04] But again, same question.

[00:34:05] So I think eventually, as with most things, Uber started, wasn't probably really legal per se.

[00:34:10] I don't know.

[00:34:11] You'd have to debate that with a lawyer, and they're still litigating all that stuff across the country.

[00:34:16] So these things happen, and then eventually the law catches up.

[00:34:18] And so the liability question will come up, but I suspect, my guess is, not a lawyer, but that at the end of the day, obviously the provider who does the care is still going to hold it.

[00:34:27] So you've got to really decide yourself what your tolerance is for it along the way.

[00:34:31] And I can see the benefit of it because it takes a lot of effort and work to actually find the studies that are relevant to the patient that you're trying to find answers for.

[00:34:40] And if you can punch in the prompt and get all that data without doing all those searching, then that would be a huge benefit to providers, right?

[00:34:49] And it can search the thousands of articles on the shoulder tendonitis and the best interventions for it so that you don't have to figure it all out yourself every time a patient comes in.

[00:35:00] And so there can be a huge benefit there, but I can definitely see the understand where a line has to be drawn when it comes to ethics and liability.

[00:35:10] Well, I just want to give you one other, you know, again, this is, I love this conversation because I think it's just a broad brush conversation, right?

[00:35:17] And so one other interesting thing is on the clinical side is what are insurances going to come back in with on the clinical side?

[00:35:24] So for example.

[00:35:25] Like, are you saying like maybe they could question your interventions based on what they're learning from AI?

[00:35:30] I think, yes.

[00:35:32] I think what they question is the same thing they always question, money and time.

[00:35:36] So my thought is, is they might come back in and say, listen, our RAI said that if you have this DX code, you get this diagnosis code across 10 million things.

[00:35:46] RAI thinks that A, it can be solved with only four visits in the auth and you probably should do these modalities in order to achieve it in four visits.

[00:35:54] And if you can't do it in four, we are going to scrutinize your clinical decisions, right?

[00:35:59] Because some of this you could say, well, they could do this years ago and they've been nipping at it, right?

[00:36:03] They always try and calculate averages, but a lot of this stuff comes down to auths and what they're willing to give you either automatically or what you maybe have to beg for, unfortunately, in some cases.

[00:36:13] And so I do think that just like tools are now reading the notes and giving you a compliance score on the clinical side, there's not a lot of leap to say, we're going to give you a clinical score and then say, okay, let's coalesce this with how do we pay the least amount to rehabilitate a patient?

[00:36:29] Which you could say, well, that's good. We want to pay less, but you can also say that's bad because you're working with the patient.

[00:36:35] You're Nathan Shields. You are a DPT. You've got all this education and it can't be done in four visits.

[00:36:41] And now you're doing the classic, you know, I'm fighting with United for an auth, which is, I mean, that's a story as old as time, right?

[00:36:47] So we'll see. Trust me on this. I'm not giving the insurance companies any ideas they don't already have.

[00:36:52] Believe me, they know what they're playing with. And so you do too. And this is where everybody's, the whole market's going to move together.

[00:36:59] And, you know, I think you'll see innovation on both sides. It's going to be an interesting tennis ball, I guess, going back and forth between the parties.

[00:37:07] Right. Definitely. Well, if people want to reach out and find out more about Empower EMR specifically or reach out to you, how can they do that?

[00:37:14] Yeah. Well, first, I always love to invite anybody I can to connect with me on LinkedIn. I'm Sharif. You can, there's not a lot of Sharif seeds on LinkedIn. So just pop my name in there and find me. But we certainly would invite you to check out Empower.

[00:37:27] We're at empoweremr.com on the web. You can read about our AI features, what we have going on. We've got some really interesting innovations always going. And we're all rehab all the time. So our practices are you.

[00:37:41] You are our clients and we appreciate that. And it's always good to be on the podcast. So yeah, please find me, find us, check us out. Love to show you a demo and take it from there.

[00:37:50] All right. Well, thanks for joining me again today. It was so good to talk to you.

[00:37:53] Yeah, it really was. Thanks for the conversation.

[00:37:55] All right. Thanks, Sharif.

[00:37:56] Bye, everybody.

[00:38:00] Thanks for joining us today in the Physical Therapy Owners Club, the resource for stability and freedom in your PT practice.

[00:38:07] Reach out and join the network today. Subscribe to our podcast, get links to social media,

[00:38:12] and access all of our episodes with show notes at ptoclub.com.