.png)
Transcript
[00:00:00]
Kelly: Hi, everyone. Welcome. We're gonna give everyone about one more minute just to see if there's any stragglers, and then we'll go ahead and get started. We appreciate everyone being on
All right. I'll go over a couple housekeeping things, um, before we get started. Um, first wanna make sure everyone we- knows that the session is being recorded. Um, and then our team will actually at Prompt will be sharing this information with the attendees, um, and any of those that signed up that weren't able to make it today.
Um, for any sort of questions that you have throughout the presentation, um, please feel free to use the Q&A, um, which is down at the bottom. It says Q&A there. Please feel free to use that. You can post questions in there. Our team at Prompt or Chris and myself will answer those questions as we can during the presentation.
If we can't get to it during the presentation, we'll go ahead and follow up with an email, um, answering those questions to everyone. So just, um, just know that we can do that. You will be muted throughout the presentation, so the questions is the best place to put anything, uh, that you might have for Chris and I.
Um, and then we'll have some folk- we'll have some sections at the ba- on the, um, at the end where we're going over, um Are where you can reach out to us and any questions or something you have may- maybe a little bit later. So please, please know that that's coming. Um, but thanks everyone for making time for us.
Um, Chris and I are excited to be here. Um, we are doing a presentation today on our therapy types, um, and what we can actually see from improvement with each specific therapy types. [00:01:30] Um, my name is Kelly Brown. Chris, we can jump to the intro page. My name is Kelly Brown. Um, I work for Prompt Health. I am the AI strategy and implementation lead.
Um, basically, what that means is I am kind of the clinical voice of our AI teams, and so I help our AI teams ensure that we have clinical relevance and that our AI tools, um, help our clinical teams and our clinical, um, um, uh, our staff and stuff, as well as our therapists that use our tools to actually be, um, helpful for them.
Um, my history is that I'm a PT by training, spent over a decade in private practice. Um, my claim to fame is I'm the first therapist to use Sidekick, um, in its ver- very earliest version. Um, so I've been part of the development of that since, um, since the beginning. So it's been, been a lot of fun and a fun journey, and I'm excited to be here today.
I'll hand it over to you, Chris.
Chris: Thanks, Kelly. So I'm Chris Hoekstra. So I'm, uh, with Hychara Health. I'm vice president of business development and operations. I'm also a therapeutic associates. My-- I'm a PT by clinical training, uh, but also I'm a clinical informaticist, so I kinda bridge the gap between clinical and business and then tech.
My area of, of research and, and business focus is really on tech adoption, uh, and really the concept of sociotechnology. So how do we understand technology meeting teams and groups and goals and those sorts of things. Uh, so I'm really heavily focused on digital health transformation and where are we going and how to [00:03:00] make sure we make good use of these tools.
So, uh, excited to be a part of this presentation again and work with the Prompt team.
Kelly: Thanks, Chris. Okay, so some of our learning objectives as we go through our time together today, um, we're gonna be looking at some common documentation types, um, that we're seeing within the data that we have for our therapists.
Um, we have this based on national analysis of therapist habits, and you'll see kind of the, the-- how we've broken that down. Um, we have an... We apply, of course, a tailored intervention to those specific therapist types, and then we're looking at what the outcome of that intervention is for those specific therapist types.
Um, we're looking at, of course, key challenges and strengths associated with each of those therapists. And then, and you know, much of this is of course, with- through an AI-driven tool that we have, um, to help support the therapists. So you'll get to see a lot of that interesting data as we move through this.
So if we go to the next section, we'll go through how we actually collected the data and what our analysis looks like. Um, so from here, what we're using is thousands of rehab therapists nationwide. These are not just physical therapists, there's also occupational therapists, a few speech therapists in there, um, but definitely a full- fully rehab therapists.
Um, we're using our Prompt Insight tool, which is a data tool, where we're able to look at the data listed below in those bullet points. So we can see compliance scores, we can see time to sign off, both the time of day that, that, that note is being signed off, as well as how many [00:04:30] hours post that visit you're seeing.
So how many hours it takes for that therapist to sign off. We're looking at their CPT code diversity, and then we're looking at their usage metrics, just generally, like how they're using their AI, um, their visit unit data, and then of course, much more. Um, so we did a great cl- cluster analysis that you'll get to see some of the visualizations for that to actually see those cohorts of therapists in our different clusters.
As we go to the next section, um, we're gonna review some of the industry trends. Um, many of you are very, very aware of this, living this day to day. Uh, we always put this in here 'cause it's important to remember kind of the framework of why we would want to do an analysis like this and why we thought it would be valuable.
If we jump to the next slide, you of course, see how we are, um, looking across disciplines and rehab therapy. We know we are, um, struggling with declined reimbursements and high demands, right? So we all feel this on a day-to-day basis. Um, we know we're grow-- we are up against a growing list of challenges, um, and we, you know, continue to have those every day.
Um, when we put this together last year, we had just had our reduction in twenty twenty-five. We're of course, knocking on the door of the CMS, um, guidelines again coming up. So we'll see how that kind of pans out. Um, but of course, we all probably are feeling the workforce shortages, the burnouts, the pressures.
Um, I find this stat quite interesting. It's something to think about when we think about the industry is projected to grow by fourteen percent in twenty twenty-three, when the average gross-- growth for most [00:06:00] occupations is four percent. Um, that's significant for us, and so we, we need to figure out a way in which we can meet that demand and of course, keep our therapists happy.
So that's part of the reason that Chris and I were very excited to, to do this presentation, um, last year, and why we're excited to be here again to present this, um, information again. Um, there's of course, the administrative burden and the patient cost, of course, that we have as well. Um, we all know that that's part of, part of what we're up against as well.
So, um, again, any technology that can help us, um, deal with all the, all the challenges we have in our industry, of course, is gonna be very, very helpful All right. So then as we go through, what we're looking at here is like, as I mentioned, part of the reason we're wanting to help with a lot of this is that, um, how can we help our therapists thrive and not just survive, right?
So we, we have this very challenging industry that we work in, but we have a lot of therapists who love what they do. They didn't become PTs because they thought this industry was gonna be the coolest thing in, you know, in sliced bread, which maybe when I started school it was. But, uh, now we know that there's a lot of challenges.
We see a lot of therapists coming out of school today who are very aware of those challenges, yet they still choose to do it because it is such an awesome profession. Um, and so we really wanna make sure that we are helping our therapists work smarter, um, and of course, do all the things with these bullet points.
But fundamentally, we want them to be able to do that while they still love their jobs, um, and that they want to actually stay in a career for a long [00:07:30] term. Um, Chris, I'll go ahead and hand it over to you.
Chris: Great. So what I'm gonna go through next is, so this concept of types of providers, and you can look at this in two frames.
My background is I'm a qualitative researcher. Uh, I'm part of a research team at Oregon Health Science University, where we study documentation and specifically scribing. Uh, and we could look at this one way, where all of you, if you look at your staffs, you could say, well, certain people type... or, or, sorry, document certain ways.
Some people are very fastidious, and they want to get everything exactly right. Some people go really fast. That'd be more of a qualitative way to look at it. What we're gonna do now is, uh, show you the analysis that Kelly talked about, where it did some cluster of factor analyses, where we looked at lots of different characteristics of providers and see how they group together and can we actually make some, um, clusters of providers based on how we see them behaving with the system So some things that we looked at, and this touches a little on what Kelly talked about, is ultimately we, we're really interested in with respect to burnout is the time of day that the providers are signing their notes off, right?
We have this concept of pajama time, where they're in their jammies at home at night when they should be hanging out with their kids and their loved ones, and they're sitting there doing documentation. We wanna watch that. Uh, we also wanna know how long it takes them to sign off the note from the time when the patient was there to when they finally hit sign and get the note out of their inbox, how long does that take?
But that relates to other things like compliance scores. If you're not aware of the Sidekick and the [00:09:00] Insights platform, the Insights platform gives kind of a note quality score. So Prompt has their version. You all might have some definition of, of, um, quality notes that you try to hold your folks accountable to.
CMS has their guidelines. So however you define it, we wanna look at those characteristics. CPT coding accuracy, so the codes you put down and what you describe and what you actually did aren't always the same thing. So at the very least in the EMR, we can look at for what we call concordance or matching between what you coded, the CPT code that you coded, and how you described it, and do those things match?
And then ultimately looking at patient adherence, drop-off rates, uh, other types of outcomes. So we try to cluster on lots of factors. What it gives you, and this is gonna be a bit of an eye chart. I know I'm talking to the Academy of Leadership and Innovation folks. There's probably lots of tech folks here, but I know not all of you are, so hang with me.
But what you're gonna see here is a graphical representation of a cluster analysis. Now, as humans, we think in three dimensions, and that's why this thing is moving around, is to show you that there's an X, a Y, and a Z axis. This is really only a graphic representation of what the cluster analysis does.
What the cluster analysis does is take many more axes than three. All the different characteristics that we've talked about, we can grade those things and put them on a multi-axis, uh, framework to then create clusters. But in the end, what it does is create these groupings, and what the analysis did [00:10:30] was these now weren't done inductively like I told you I do in my research.
These are, the data told us these are characteristics that hang together. And then as a research team, we came back and looked at these and said, "Okay, how would we label that based on what it says?" So for instance, you see here the traditionalist in, in this purple one. These are people that tend to be pretty conscientious.
They tend to have good compliance. They do what you told them to do. They're probably a little more verbose in their notes than other folks, but they, they get it down, uh, and they generally code pretty well. By contrast, the purple would be a procrastinator. These people don't necessarily have a ton of extra patients.
It's not like they're seeing more patients than everyone else, but they just take a long time to sign their notes off. They're just pushing it off, not because they're seeing more patients, but just because they're not getting it, getting to it Another one we'll talk about is the habitual over-coder or the autopilot.
These are folks that, uh, just kind of, they know they're supposed to code a certain way, so they just throw it down. Their notes are usually pretty sparse. Uh, they're pretty good because they're gonna give you lots of codes, but their notes don't support it. So I'm telling you all of this because we're gonna go through not all of these, but a few of these groups that cluster together and try to define them, and then ultimately that'll walk across to how Kelly's gonna finish with us, is, okay, well, if you have this type of individual, what are the things that they need help with, and how might the system actually improve their lives and their coding?
So again, just in now two dimensions, so the thing [00:12:00] stops moving, here's where we're gonna focus for the next, next section. Now, before I dig into these actual types, uh, archetypes of documenters, we need to recognize that when we do a factor or cluster analysis, we get this nice little blob, and it looks like everything hangs together really cleanly.
There's a lot more noise in the data than what you see here. And so this is an example of, we took a, a few of those clusters. There's four of those clusters here, and then we said, "Based on those clusters, each of these dots is a provider, and the size of the dots is the number of cases that they have or encounters that they have."
And so we took these clusters and we said, "Let's take a look at how these folks are coding by functional code." So if you've heard anything from the Prompt group, we're really talking about these high-value functional codes, their activities, neuromuscular re-ed. Uh, how are they billing those as a percentage of their total units versus, are, are they high?
Is it a hundred percent of the things they're coding? Is it low, zero percent? We want it to be in this green to orange section in the forty to sixty-ish range. And you can see by and large, most people do this well. What we want to point out here is that, and this will draw it a little closer, in the middle, in this red section here, if you were just take the average of your entire organization, you have a ton of the greens and the oranges, and you'd probably say as an org, "We're doing pretty well."
That, uh, most folks are in the forty to sixty, and so we're not too concerned. But what we need to show you is there are hidden between those people that fit the norm, [00:13:30] these outliers, the reds, they're there. There's actually a fair number of them, but they're hidden. And if all you did was look at your organizational averages, you would lose sight of the fact you actually have some people that are a risk to your organization that you need to deal with.
So moral of the story is, one, don't just look at averages as a whole. Two, even when you're looking at just these individual archetypes, you're gonna have some edge cases that don't fit the norm, and so you're gonna have to look a little individually. These are all overcoders, the folks in the red. Or overcoders, we would say, are using more of those functional codes, high-value functional codes than other folks.
It's not a bad thing. It's just you need to know you're an audit risk from a payer if you look different than the average. But also, you have these cold spots, these people that are way underutilizers of those codes. So even though your average for the company looks good, you're gonna have some people, a spattering of them, but more than a few, that you still need to deal with.
So moral of the story is you're not gonna have one intervention for all of your, your providers. You're gonna have to tailor your intervention for the providers that you're working with and their specific issues. So we can look at this a slightly different way is this is now five of the different clusters, and we then said, "Well, within these clusters, let's look at time to sign off."
And this just looks like a confetti bomb. So there actually is no consistency or what we call homogeneity here, that inside of any given cluster, there's a ton of people that take, uh, well over an hour to sign their notes, multiple hours to sign off, and those that can [00:15:00] do it really fast. There's no real rhyme or reason to these.
So the cluster analysis doesn't always give us something useful. Or said another way, if I looked at this, where do you start? 'Cause you have, in any given cluster, some people that are taking way too long and some people that are doing it really fast, and everything in between. There's really not, again, rhyme or reason If we looked at compliance, so again, this is a few clusters put together.
You don't need to worry about which clusters they are at this point. But even within these clusters, we have folks that are really compliant with their notes. So based on the algorithm that Prompt uses, they're writing quality notes. That's what the green folks would be. Probably green is a little too quality, but we could debate that another day.
And then you get people, uh, down into the orange. There's not a lot of red, so most people across all the clusters, you have good and bad compliance, um, across them all. So I showed you that because we just need to recognize as we now start to talk about each of these archetypes, you need to recognize that this is an average.
We're gonna talk about the prototypical this person. You probably have the person in your mind when we describe these things and say, "Oh, yeah, I can think of a person that, that documents like that." But the pr- the past few slides need to tell you that there's a lot of blur around the outside, and there's not a...
No one fits exactly into this box, or two people that are in this box might be slightly different from one another on different domains as you measure them. But neither here nor there. Now we'll get to the so what [00:16:30] of this all. So the traditionalist. This is a, a person that is highly compliant, so they're doing what you want them to do.
Their notes, they're fastidious with their notes, so they're detail-oriented. They're typically because of that slower to sign off than would be ideal. They're probably prone to burnout. They're pretty rigid in their coding habits, so if you wanted them to code a certain way, they're gonna code that way because they say, "I follow rules."
Um, it's, it's really that they'd be your poster child of the compliant child. Those of you who are parents, some of you have compliant children, and some of you, like me, have non-compliant children. I have three, and they're all different. This would be the very compliant parent's dream child. So high compliance scores, low time to sign off, low CPT coding diversity because they're just-- they found the path how you're supposed to do it, and they do it.
The good of this is they're pretty conscientious, and they have a very high standard. The bad of this is to get them to change their behavior, pretty hard, because they are locked into process, and they don't like giving up control. These are also the folks that want their notes to be perfect, and so really to satisfy those people is gonna be really tough because they have high standards.
And then because of this, even if they have a great AI tool like Sidekick, they may not trust it. And in our research, in our qualitative research, we find lots of folks that would say, "Your tool might be great, but it's never as good as me, so I'm not gonna use it." The tool can't help someone who's not gonna use it.
And so we just need to think about it, this person needs to be addressed a certain way. Now, back to the [00:18:00] variability. This, these are all traditionalists. I don't wanna get too hung up in these numbers. I wanna get to Kelly's part, but I wanna show you that there's variability. So if we said this traditionalist is characterized by higher than average compliance, we can see of all these folks, there were 8,000 of the, of the providers that met that standard.
They had longer notes than other folks, about 6,000 or 6,800 fit that. They had high average time to sign off, about 7,300, and only 1,700 had all of those factors together. So we just need to recognize, again, there's not homogeneity here. I don't wanna get you too into box and whiskers plots, but basically if this is all of the users of the system, and these are those kind of most descriptive of the traditionalist, how do they set?
You can see by and large a box and whiskers chart. The orange line is the median, the diamond is the mean. This is the maximum, and this is the minimum. You can see there's a ton of variability in the overall data, but these people that are most prototypical of the traditionalist with respect to compliance, they live up here, uh, quite a bit higher than the average of everybody else.
Not quite as, uh, um, uh, significant when you look at the average number of words, and then it's ... There are a lot of variability when you look at time to sign off. I'm not gonna spend this much time on all of the slides. We're gonna go through each of the archetypes. But just to recognize there's a ton of people that have at least one of these characteristics.
Very few of them have all of the characteristics, [00:19:30] and then within them there's a lot of variability, but we can generally see a trend that these folks take long... or sorry, are better compliance with their notes Longer notes and higher time to sign off. If you guys are more graph people, this shows you of all the users, this is where those most prototypical people for average compliance sit is way up here in this top quartile.
Um, same thing with respect to average words. And then there's a lot of variability here. This is a really skewed chart. Uh, not a lot of people, but average time to sign off is there. The next, uh, archetype we talk about is gonna be the habitual overcoder. So this is one where the therapist coding intensity doesn't necessarily match their justification.
They might be really good on billing those high-value codes, their activities neuromuscular re-ed, but they don't justify it well, so there's a discordance between how they describe what they did and what they coded. And they might not even be aware of that compliance risk. We generally see with these folks, they have pretty broad, uh, CPT coding diversity.
Their compliance, their notes are n- generally not great, so they're probably just coding well. They're making you lots of money, but they're not taking a lot of time in their note to describe well what they do, and that shows up as a lack of connection between their coding and how they describe it. So there's a lot of disagreement between what the AI CPT suggests they should code based on what they wrote and what they actually picked for their code.
The pros of this person is they're making you a lot of money, that they probably have learned somewhere, "I need to do this." They can bill high RVU things. They're pretty [00:21:00] operationally focused. They are a great opportunity for AI support because they probably don't even know that they're a risk, and they wanna find a fa- a fast path.
AI can certainly do that for them. The cons for these folks is, again, like I said, they're an audit risk for you, and they're probably a creature of habit. AI is a good tool, especially to get in their workflow to help break their habits. Again, I won't hang up on here. There's a lot more smushed together.
There's not a big difference in their CPT per units that are billed from the average. There is a pretty big difference relative to their time to sign off, so this says they're just getting through their notes and hammering through and signing them off. Um, but it's probably not the best note. And then ultimately, they bill slightly more units than everybody else.
Okay, and again, you can see in the graphs the same thing, kind of where they sit in these tails. Okay, the autopilot one would be the next one. This is a therapist that just has learned a path forward, and this person is just, they know how to do it, they can get it done fairly efficiently, and they just have a well-worn process.
But they don't engage well with their notes. So you're gonna see really great sign-off time, but pretty short notes, because they just wanna get in and get out like the last one. They typically are pretty low on their functional units, because they're like, "I just did what I did. I put it down. I don't need to differentiate between their activities and neuromuscular yet.
I just put it in." Their compliance is pretty low on their quality of their note scores. The pros to this is they're focused on their patients. The, the documentation is a side thing. AI allows them to set their computer behind them, but they [00:22:30] wanna just hammer through and get stuff done fast, but they're engaged with their patients.
So make sure that you emphasize the positives. Like with a couple of the other ones, though, they're probably an audit risk for you because they- they're a little more flippant in how they do their billing, and they don't justify it well in how they describe it, and then their coding is generally poor, documentation is poor quality as well.
But it's fast Again, same thing here, pretty smushed together. You don't see a big difference on their number of words. You don't see a huge difference on their compliance scores. They are pretty clumped down here, though, with respect to their, uh, units aligning with the CPT insights, and we can see that there's a lot of them lined up over here in the lower quartiles, meaning they don't line up well.
Okay. And then last one, the treadmill therapist. This, this therapist is, they want to do the right thing, but they're on the daily grind, that they ultimately are getting in and getting out. They're getting-- They have high patient volumes, um, but they're just stuck in these cycles, and they can't catch up.
Unlike the first one we talked about, the procrastinator, where they don't have a ton of extra patients, they're just taking forever to do their notes because they're distracted or they don't prioritize it, this person, we would infer, can't get to their notes because they're seeing a lot more patients than average.
That's the first point here. Because of that, they have a high late note-- or, sorry, late-night sign-off. These folks are gonna have a really high patient drop-off risk because they are just on a treadmill just [00:24:00] trying to make it through the day white-knuckled and then sliding into bed at night and falling asleep after you do an hour's worth of charting.
These people, one, are a burnout risk, and when you're at a burnout risk, you're probably gonna have patients dropping off because they just... You're not gonna have a great experience. Pro for these folks, it's kind of a, an odd pro, but they are an awesome opportunity for AI assistance because they're trying to do the right thing.
They're stuck. They just don't have a path forward. They will be a perfect adopter for this because they're gonna see the value straight away. Uh, now, the risk for this person, and we see this in our research, is this person very quickly can become dependent on AI because you've shown them the light. They now can spend time with their kids.
They now can get out and engage with their patients. And now maybe they overtrust the AI. They don't go back and actually change anything. They just fall back into another habit that in some respects is better because it's getting them through their day faster, but now they just ha-have the habit of just accepting AI every time.
So again, what we're trying to do with all of these is help you to see these different archetypes, knowing that there's blur and some heterogeneity within a group, but they have different needs. They're gonna adopt technology in a different way, and then you're gonna have to measure some things for them with use of technology to make sure that that technology is having a positive impact on them and their impact on your organization.
And again, these are the graphs. You've seen these. I don't wanna harp on, harp on those. So now that's kind of the archetypes. Now the real meat of the c-talk is gonna be Kelly talking about how the tool actually helps these folks [00:25:30] individually.
Kelly: Awesome. Thank you, Chris. So now what we're gonna go into is the intervention, right?
So we have the, um, archetypes that we've, you know, as Chris said, that the data told us about, that the data was telling us about. Um, as Chris said, we're very-- much of us, many of us are very familiar with those types of therapists. We might identify with one or two of them, um, ourselves. But now we are seeing, like, what is the intervention that actually is going to help these therapists, and what's the outcome of the intervention?
So if we go to the next slide, we'll actually highlight what the intervention is, which of course is an AI scribing tool. Um, and so I'm, I'm assuming with this group, most of us are familiar with that technology, but I will go over exactly how this is going to work. Um, an AI scribing tool is going to be an ambient listening tool that's going to give you real-time intervention, where it's recording your conversation with your patient and then, of course, writing a compliant note.
Um, the tool we're talking about here is our Sidekick tool within Prompt, so Prompt Sidekick. That specific AI scribing tool is built with the backbone of our compliance models that we actually started our company with when it-- into the rehab space. So that's where we get into a compliance-specific type AI scribing tool.
Um, we also have a component of it, as Chris mentioned, where there's some CPT coding suggestions. And so it's looking at the concordance between what code was selected and billed versus what the actual intervention and justification says, and does that match? And so-- and if it doesn't, then that's where we actually give that [00:27:00] feedback in real time to the therapist to, again, which that, that third bullet point, the provider retains the final say to actually change that.
So if they look at that and they say, "Oh yeah, I should have billed therapeutic activity. I have all sorts of functional information that I'm saying there," um, then they would change that code Again, down the, the last one I'll highlight is that this specific AI scribing tool, Sidekick, we have trained on specific clinician discipline.
Um, so first hierarchy, of course, is rehab discipline. Um, you wouldn't want to apply a physician, um, scribing tool to use in PT, um, but a PT, um, scribing tool and then of course more specifics as you become a different discipline, of course, in the rehab space, as well as specific specialties too. When we go to the next slide, we actually highlight the rollout here because this is important.
And for those of you... I'll just give a quick plug. For those of you going to PPS, Chris and I are presenting a lot on this particular section, um, coming up for PPS in, um, Colorado Springs. So we hope to see you there. Um, we have a great presentation that we're excited to present there. But here, what we see is for all of these specific users, we have the training that they went through.
So we have virtual training, support documentation that they had access to, in-app support, so the actual tool itself offers, um, in-app support in different ways, and then of course the support team through Prompt Health that we have. Now, one thing we aren't accounting for here, which is gonna be a lot of what we talk about in our PPS presentation coming up, [00:28:30] is what sort of support the actual clinics and, um, companies implementing this are s- are providing.
Uh, we'll talk a little bit about that later on, um, but that we, we aren't accounting for that here because that's not something that was, um, consistent across the board, um, but these things were. Additionally, we do have our adoption monitoring. So this is where we're actually seeing like who is actually using the AI scribing and how.
And so here's where we can see the frequency of usage. We can see the audio mode. So there's a couple of different ways in which a therapist can use the scribing tool. Um, there's an what we call auto, um, auto generation, where you just like record a full length, you know, conversation with your patient and it writes all the content of the note as you need, versus a more specific what we would call like a smart dictation mode, which is more of, more of you who are maybe used to doing like a dictaphone, where you would say, you know, "Subjective," and then you would say the subjective.
What makes it a smart dictation is the AI is gonna clean up the verbiage. So it's gonna clean it up a little bit. It's gonna take out "ums." You don't have to be quite as specific as we used to when we were using dictaphones. Um, but those are things that we actually can look at when we're looking at adoption monitoring And then, of course, what we're going to look at now in the next upcoming slides is our before and after metrics, um, in line with what we talked about with Chris, Chris just went over.
Chris: So if we jump to the next-
And the o- the only piece I'll add here is not being... So I'm not with the Prompt team. We're a user of [00:30:00] the Prompt tools, and I'm a, an informaticist that kind of interested in tech adoption. I, I can't emphasize enough what Kelly said, is that the, the organization taking ownership to augment the training you get from the vendor is imperative to getting a good outcome.
Mm-hmm. One, just to get your folks to, to adopt the tool, but two, to get a good outcome. And then the other piece is monitoring. You, you can't necessarily... If you just looked at the back end and say, "Did it improve my coding?" You could look at that as an operator. But if you're not looking at are your people actually using the tool or how they're using the tool, you can't design intervention.
So you need to monitor things, one, and then you need to own what are you trying to get out of the tools. The vendors... Now I'm getting into our talk for PPS, but as you look at this, you're gonna see the next few slides where you actually see some change. This is from some orgs that were highly engaged, plus some orgs that were not.
I contend that if we were to take a look at those orgs that are highly engaged and pull them out separately, you'd see a lot bigger difference because that's so important to getting a good output or outcome from, uh, from tech implementation.
Kelly: Yeah, for sure. Um- Okay, so then let-- I wanna orient you to what we're looking at on this particular page.
So we're gonna go through each of the four archetypes. Um, this one first is the traditionalist. So here we have three separate categories that we're looking at. We're looking at words per note, we're looking at hours to sign off, and then average compliance. The gray bars, so the light gray and the dark gray, those are all Sidekick users, [00:31:30] both pre-Sidekick, so three months, at least three months prior to using Sidekick, and then the dark gray is three months post using Sidekick.
So you can see changes there. And then when we look at the blue here, this is where we actually see are the traditionalists pre and post. So some of the things we're seeing off here, and I'll kind of, um, highlight this in the order of the, um, the bullet points, is we see a faster sign-off. So that's the, the, the middle graph here.
So this is where we're actually seeing a seventeen point four, um, hour reduction in their hours to sign off. And so this is again, from the date of service or from the time they're seeing the patients until they sign the note, they're actually cutting that down quite a bit. Um, but what we then see is we actually don't see a decrease in their compliance, which means the quality, the way we're measuring quality with our Prompt, um, insights, is that we actually are seeing very, very little change in their quality.
So they're writing a note faster, they're maintaining the quality of the note, but then in that first graph here, we're actually seeing that they're cutting down on the content. And so that's where we see these are the therapists who just, they feel like they have to write a whole story in order to get everything in a note when we know that's not true.
When we think about from an auditor's perspective or from compliance, we know it doesn't have to be super, super verbose. It needs to be good documentation, and that's where we can see we maintain good compliance, but we [00:33:00] actually see a reduction in the number of words per note, um, which actually, of course, then helps people be a lot faster there.
And then of course, you have less to read if you're looking back on that, which of course is very, very helpful here.
Chris: And, and this is paradoxical as a operator that works in an organization because Kelly would say, "Well, their, their compliance didn't get that much worse." But you need to couch that against to say they are already dramatically better than the average PT.
These are the people that are the A students, and if we teach our folks anything, it's like you do not have to write a hundred percent perfectly correct note. Like, that's not our intent. Our intent is a good enough note that's not gonna get you audited and it's gonna get you paid. So these folks are learning that I don't have to be 100%, which is, it's fine to let it drop because they're still so much better than the average.
Kelly: Mm-hmm. And that's where from a training perspective on our side, we actually try to train our folks, to Chris's point, where we want our therapists to hit about an 80, 'cause that's where you're, like, right in that green realm. You're not too, too much information. We know you're not putting too much information, but we're right on that.
So Chris, that's a good point of, like, they, they probably could afford to lose a few points of compliance score because they just have too much in there. Like, it's just unnecessary information. Um, but yeah, that's a good point Oh, and actually I wanted to-- and, and it's relevant on this slide too, but, but one thing, Chris, that you said in-- just a moment ago is that the ongoing, um, monitoring of usage, this is also important when we're talking about the outcome too, [00:34:30] right?
So this is just three months of using Sidekick that we're looking at here. But you can imagine this therapist as they become more comfortable with this particular technology, as they understand how it's helping them, as they become more and more comfortable. Like, there is a very fast learning curve with this technology, and we do see that because we see a drastic improvement and change within three months.
But there is, you know, a continuation of how do these therapists actually adopt it into different variable, you know, different types of patients that they're seeing or kind of the outliers that they run into, and then how can they really use it for all of the different things that they see on a day-to-day basis.
And that's where we would anticipate if we were able to do this for a further amount of time, that this particular type of decrease, you know, like more appropriate compliance score, their hours to sign off, their words per note, would continue to be more in line with what we would expect to be normal. So-- and that is gonna apply across all of the other ones that we look at here too.
So then now similar setup, but now we're looking at the habitual over-coder. And so again, if we kinda go from left to right on the specific, um, metric we're looking at, we have our hours to sign off, we have our CPT units per visit, and then we have our functional units. So the story that we can kinda see with this particular data is that we actually are having in that functional units there, we're having where there's an improvement in compliance with the CPT guidance.
So many of our operators are going to look at that, that, um, right-hand section where they're seeing a decrease in the functional [00:36:00] units and go, "Uh, that's less money for my organization. This is not good." But if you remember back to how Chris highlighted this particular therapist, this is the therapist that also puts you most at risk from a compliance perspective because they are not selecting the appropriate code.
We see that here in the data where the tool itself is actually giving them the feedback that they're over-coding. They are selecting a functional code when they do not have the justification for it. And here we can see it actually is reduced. Again, they're significantly higher than the rest of the Sidekick groups.
So just remember that. Like, all the rest of the Sidekick users, they're still significantly higher, but we've slightly edged down, which tells me that they're likely much more compliant when it comes to actually selecting the appropriate code. So that's the story that we see here But if we go to the middle graph where we're looking at the CPT units per visit, they are basically the same when it comes to the number of units they're billing.
So that's a good, that's a good thing. And then of course, their hours to sign off, we see a slight increase in their hours to sign off, which again, because of how drastically different they are than the rest of the Sidekick groups or the rest of the Sidekick users, um, it's not necessarily a bad thing to have a slight increase.
Again, they're still under ten hours. They're still, you know, getting those notes signed off within the same day of seeing that patient. Um, but now they have more feedback. Are, are they collecting or are they charging for the right code that they've actually documented? So again, [00:37:30] something that, um, some of our operators look at this and go, "Ugh."
But we do know that it's very, very important to justify for what we're coding for. That, of course, is over the long term, something that we wanna make sure we're not getting audited on and having to pay back those types of things. So of course, for long-term ROI, you wanna, wanna see this as well We jump to the next slide.
Now we dive into the post Sidekick AI scribing tool for the autopilot. Again, to orient you to the metrics we're looking at for this particular archetype, we have our words per note, um, we have our average compliance, and then we have our units aligned with the CPT insights. And so here's where we're looking at, again, the middle, middle one here with compliance.
We're seeing about a sixteen-point improvement in compliance. That's just with using Sidekick for three months. Um, again, that's where you know that therapist needs that assistance. They're a compliance risk for your organization, so now Sidekick helps them do that. Um, we do see that alignment over on that right-hand side there, the alignment in the CPT insights increases over ten percent.
What that tells us, and that's where in that second bullet point there, there's a boost in CPT diversity, which generally leads to higher reimbursement potential. Now, the reason that is, is because we've done-- Chris and I have done a couple other presentations together where we've analyzed how do therapists actually distribute their CPT coding.
And many of you are probably aware that there is an overutilization of lower-value codes, so therapeutic exercise, and [00:39:00] an underutilization of therapeutic activities and, and neuromuscular reeducation. So what we know is that if we can align our CPT insights and have greater concordance with our AI suggestions, uh, with our actual what we're doing, then what we actually see is that there is an increase in that CPT insights, that CPT alignment, and normally we have an increase in our functional utilization as well.
The, the highest, um, the most common thing we see when it comes to coding malalignment is that therapists are actually doing functional exercises, but they're not coding for it. They're doing therapeutic e-exercise. They're billing therapeutic exercise instead, when they should be using therapeutic activity.
So that's what we know to be true here. Then, of course, we see an increase in the words per note, um, in that first section there. The thing to highlight here is that that is not an added burden. And so when we were typing notes or we were handwriting notes, um, that would have taken us longer to do the note because we would, of course, had actually had to write that.
And so when we don't have to write it because the scribing tool is doing it for us, we have increase in the words per note, which you can still see is significantly less than the, um, most Sidekick users here. Um, but we know that the compliance goes up. We know that the coding alignment goes up, so-- but it's not an added burden.
So that's where the scribing tool can really, really help this specific group.
Chris: And from the user, from a kind of an operator in an organization, this for me [00:40:30] has been the most compelling slide, 'cause this is the old dog, new trick slide. That most of us know this. I was this person prior to using Scribe. But you just found your path, you've been doing it a long time, you know how to get through, you know it's not gonna get denied.
You just get en route and you do it. The fact that in three months you can implement technology, and these little tails are the error in this, just see how much change there was compared to the error. This is significant, sig- very significant, the change in only three months. And again, th- keep in mind, these are people that are ingrained in a habit, and you've changed them in three months.
Uh, this is probably the, the biggest example of where the technology can move.
Kelly: Mm-hmm. Yeah. So then when we jump to our final one, so here's where we're hitting on our final therapist, which is the treadmill therapist. So here we have four metrics that we're looking at. So we have our hours to sign off, we have our average compliance, and then what we did with our, um, sign-off metrics there is we actually, um, separated them out into two different categories.
We have our after hours, which is kind of like after the clinic's closed, they may stay late, do their documentation. And then there's the late night sign-off, which is going to be, like, into the, you know, late night or wee hours of the morning. Definitely pajama time, as Chris calls it. Um, and so what here as we're seeing is that there's an improvement in the time to sign off in general.
So you can see about in 28.3 hours, so you have a day's improvement in the time to sign off. So that's in that first section. So for those of us who are [00:42:00] conscientious about how quickly we're getting paid, you of course have to sign your documentation quickly. So here we have a day's improvement for these particular therapists.
Um, we do see a significant improvement in compliance. Again, we're seeing a 12.5% improvement in compliance. We're now getting up into that 70, so we're matching up with most of the other Sidekick users. And then of course, we see a decrease in the time to sign off in that late night. So both in the immediate after hours, like after work, and of course in that late night sign-off too.
Now, there's still some habitual, like, there's still, like, I, you know, we have some Sidekick folks that we work with, and they just like to do their notes then. That's just when they, they ha- It's just what they like to do. And you know, I, I had a, a s- period of my life where I would wake up early on a Saturday and I would sign my notes, and that's just kind of what I did.
It was almost like a routine, and, uh, something was really great about that. Now, you can't always break all of that, but we can at least see that there is a significant improvement in this. Now, if there's a therapist who really wanted that to go away, then they would have the tool to do that, and this is that therapist who'd be able to make that happen with the tool.
So again, when we talk about it, like, to me, I see this one and I'm like, this is a golden, this is golden therapist, type of therapist to use Sidekick and get a lot of help. Um, and so I love this particular graph too
Okay, so then that was the overview of the pre and post, um, using Sidekick. Now we're gonna get into some kind of, um, reports from our clinicians using [00:43:30] this. Christine, she's a therapist that works with Chris at TAI. Um, her story is very impactful. Um, it's wonderful. I love talking with her, um, about this.
And so she of course, has been a PT for thirty-four years, more now. Um, she had resigned with no plan of returning. She had heard about Sidekick, um, and Prompt Insight, and she had heard about the technology, and she actually came out of retirement and went back into practice and working for TAI in other aspects, um, because she was so moved at the way the technology is actually making a difference in our industry now.
Um, and I'm not, not putting words in Christine's mouth, but her and I have talked about, like there's been a lot of technologies that have come, come and gone throughout the years, um, but scribing is the one that has had one of the biggest impacts and when it-- actually lifting the documentation burden off of us.
Um, if we go to the next slide, we'll see a couple more quotes here from other therapists. Um, these ones I'll let everyone kind of read through as you go through this. Um, one of my-- I mean, we have everything from how long it takes me to do things to, you know, Samantha with that first one, that her kids thought she quit treating patients 'cause she wasn't doing notes at home.
Um, that to me is like when we talk about quality of life and, um, you know, our children just think we write documentation versus what we do is like see patients. And so, um, I thought that this, that was very meaningful having children myself. Um, but there's a lot of good, good information here, which we love As we go to the final slide here.
Um, so [00:45:00] as, as we mentioned, Chris and I presented this almost a year ago. Um, since then, our Prompt team has actually taken this information and made a subjective quiz that anyone can take online to actually determine which of the four archetypes you might be. And so please everyone go ahead and re- um, take a screenshot of this.
Um, you'll actually be able to take that quiz. When you get to the end of that quiz, there's some, um, mentoring or some kind of coaching things. Myself and Pedro, who Pedro is one of our founders who, um, designed Sidekick, he and I actually go through a coaching session for those who are each archetype on what to think about when it comes to using the tool.
And then if you are a manager of folks who have multiple different types and you have multiple archetypes in your clinic, what you would do with each one of those. So a lot of fun to kinda go through. Um, well-made from the Prompt team, um, but a lot of fun kind of, kinda to understand this. So please, please, we tried to make-- take the information that Chris and I pulled out in this, um, presentation and then, um, put it into something useful so that we could actually take that and put it into practice when it comes to our, our own, um, clinics.
Um, and then that is the end of our presentation, just in time to take any questions. Um, we do have our email addresses here. You can feel free to reach out to Chris and myself, or if you have any questions about Prompt in general, you can reach out to support@prompthealth.com as well [00:46:30] ahead and take a look at the questions Yeah.
So for, um, Ian, you a- asked where, where are the mentors videos based on archetypes. They're actually, once you go through the, um, quiz, you'll get to the end, and that's where you'll, you'll see the mentoring videos. So you gotta go through the quiz there
Chris: Mm-hmm ... seconds to decline for the treadmill therapist. Someone asking a stats question. You're my people
Kelly: I
Chris: know ... second one. All right, there we go. And Patricia, is this the one you were hoping for?
And let me know if you have any specific questions off of this one. But the key with all of these is, uh, you can see these green, um, uh, chart, uh, distributions down here are really small. There's not a lot of therapists that fit in each one, and they're spread out over this top quartile Oh, okay. Patricia was here for the stats, so feel free to send me any emails you want- Yeah
and we can geek out on confidence intervals or quartiles. Yeah, that'd be great. I don't get to do this very often. All right. Uh, anybody else?
Kelly: Uh, so Ian, you're asking if [00:48:00] Prompt and Sidekick do an education session for clinic staff. So yeah, that's part of the training that we do. Um, so our, our clients, client success managers, they do support all of our clients for, for Sidekick. So they would do an initial training. Um, they would... And it, it really is how, how your kind of organization needs to be supported.
And as Chris said, like there's, there's a lot of variety in the way in which people, um, are trying to support their clinicians and implement a new technology. Um, and so we as a vendor try to take on some of that where we do the training. We show, like, how does the tool work, what do we need to do, um, and then of course, we're there for questions too.
So, and then any sort of like in... like up- update in our, a new feature or anything like that, we're of course communicating to our, our clinicians as well.
Chris: And I'll echo what, uh, what Kelly said, that they're, they're great about adding support, and I'll come back to where I started. Coming from the provider side using Prompt, I can't convey enough that it- it's actually incumbent on the organization to, to start that by just being clear what are you trying to accomplish with the technology.
So just help your staff be aligned on that. And then from that, find your champion. So the Chris Panagos that, um, that Kelly showed, she actually became our champion. She trained the entire company on the tool. And when you can find someone who's a maven and find value in it, and then have them kind of present to the masses on how this is helpful, it's gonna have a [00:49:30] lot better buy-in than just ha- asking the vendor to come in and train more.
And so it's being clear on what you're trying to do, use your people to convey that message, and then certainly have the vendor there to help you. Um, all three of those are gonna be essential to get where you wanna go. Mm-hmm. Anybody else? Any other questions? Otherwise, we're happy to stick around after.
As Kelly said, we have our emails here. Mm-hmm
Uh, anything particular underutilized? I'll give mine from the provider and I think Kelly's- Okay ... seen a lot more folks. Yeah. Um, I will say, um, sheepishly, uh, we used the tool, and we were a very early adopter. We helped them kind of building their model out very early in the process. Uh, but we used the tool probably for almost 18 months just saying, "We're just gonna use these pieces and we're gonna throw pieces out, we're gonna put some..."
No, maybe it was 12 months, but we, we really, mm, eased ourself into it. I think they've refined the process now. I think you can do it much faster. So I, I think the, the underutilized piece is just getting it out to the masses and let them use it. Um, and if I speak to the, the, the cousin to the Sidekick tool is the insights.
Mm-hmm. That's the one the most underutilized. I think everyone can turn on the scribe and everyone sees the value [00:51:00] tomorrow, but actually using the data that comes in the insights, we used that for two years before we really rolled it out to the company and had big strong initiatives around it. We wanted...
We didn't trust the data. We went back and forth to refine the data. It ended up being all fine, but that needs to be a key of what you're doing 'cause you wanna show the value of the tool to your stakeholders, your investors, your owners, your executive, but also to help your providers better use the tool, you have to use the insights tool.
So for me, that's the one that's, even in our org now, and we use it a ton, is underutilized compared to the Sidekick.
Kelly: Yeah. No, that's a good point. I would say, well, we... I'll, I'll speak at a clinician level specific to Sidekick. One thing that we, with, specific with AI scribes that I see underutilized all the time is that everyone immediately understands how to use a scribe with an eval So they're like, "I'm gonna use this eval, and this is gonna be awesome.
It's gonna go from taking me twenty minutes to taking me five minutes. This is gonna be awesome." That is like the most immediate aha moment. People are really excited about it, but then they never kind of go further in, and that's where like the extra adoption, the metrics and stuff. And so they don't then do this like, "Well, how can I use it for a daily visit?
How can I use it for, um, a progress note, a re-eval?" And Chris and I have talked a lot about this because scri- the thing, the-- some of the magic behind scribing is it forces you to [00:52:30] verbalize out loud what we typically keep internalized and we keep in our brain so that we don't lose it. I always think of it like leaks out my ears before I put it on my documentation.
And so instead we... And this is what some of the training that we try to do is like talk to your clinicians about verbalizing to your patient what is actually going to happen. So like for a progress note, for instance, like if I'm seeing Chris as a patient, I'm saying, "Hey, Chris, since your initial evaluation, how have you been feeling?"
And then we go through that as subjective, right? That's pretty easy to capture. But then when we get to the assessment portion, I'm actually telling him, "Chris, I see that you've actually improved in this goal and in this goal and in this goal. Here's another goal that we actually haven't made much improvement in on, and I think here's why.
This is how I'd like to adjust what we're gonna be doing with you. This is what we're gonna focus on next time, and here's how we're actually going to move along with our care." That is a by far better assessment than keeping that to ourselves and then writing it in the note. And so that's one of the things I wish more people utilized because they just think that AI scribing is a time saver, but it's not.
It's a quality improvement from a compliance standpoint, but it's also a quality improvement from a rapport standpoint in helping you to actually prove value to your patients of why they should be there to see you. So that's one of the things. I could do an entire hour-long conversation about that because that's like one of my biggest things there
Chris: Yeah, [00:54:00] I- so I'll echo this.
This is a... We do a cohort with our PTs every quarter. We bring them together and help with different practice optimization. And, and I challenge folks with this. I ask them, "How many of you have signed off an evaluation, uh, in w- at least one evaluation per day before the patient walks out the door?" And actually, six months ago when we chall- when we did this, people were like, "What?
You can sign an eval off before the patient walks out the door?" I just did this about three weeks ago, and over half the class said, "Oh yeah, I, I sign all my evals off before the patient walks out the door." You can only do that if you have that kind of conversation that Kelly talks about. That most folks will say, "Oh sure, I can do my subjective with that, with the narrative.
I'll have it listen to me when I'm doing my objective, and then afterwards I'll come back and either do my dictation of my assessment when I'm sitting at my desk, uh, or I'll write that myself." You've missed a beautiful opportunity to have an amazing conversation with your patient, and then have that codified in your note.
And oh, by the way, I sign my note off and go have a drink, high-five my aide, and go see my next patient. So you're missing out if you're not using it for that.
Katie: Yeah. Yeah, for sure.
Chris: Um, Alexis did put a question into the Q&A. So she had said she uses a different scribing tool, uh, for most of this, but gets caught in the weeds as a traditionalist and really wants to optimize her coding, uh, and use that, uh, uh, to support tech as a tech adopter, but not feeling like she's getting the results she wants.
I think that the best thing for this group that we did is, is we, we were very dogmatic in therapeutic associates when we rolled this out to say, [00:55:30] "What are the elements of a good note?" We took, unpacked what was the insights and said, "What are the elements of a good note?" And then asked people, like, "What are you afraid of?"
And most people said, "I'm afraid..." This is verbatim almost. "I'm afraid that if I get called into to court on my note, that a lawyer's gonna make me look stupid, so I have to write everything down." And our retort has always been, "That's their job to make you look stupid. And actually, the more you write, the more fodder you give them to make you look stupid."
So freeing them to say, as long as it achieves what you want, which is the intent of what you did, it accurately describes what you did, and you can justify that, and then you can use it to guide your next treatment, sign it off. Like it's, that's what it's used for. So really ask yourself, why do you feel you need to, to do more, and what's the objective?
And ask yourself, "How can I achieve that objective by writing less?" Um, most people are just caught up in more is better, and that's not always true.
Ja, til de bare i jaks
All right. Well, we love getting emails, so feel free to, uh, to, uh, write to us. Yeah, and Taylor, thanks for the response. Please attend our PPS presentation. Yeah. They didn't put us on Saturday this time, so it's not on the last day, so you can all attend and not be on flights. We're looking forward to seeing y'all there.
Kelly: Yeah. Thanks everyone for making time for us.[00:57:00]
I'm still here
Pre-designate CPT codes in an eval, like set them for a specific payer or set them for, like, a specific, um
I think you'll have to reach out to your EMR support on that, so your CSM support, Patricia. That's a good question. From a Sidekick perspective, we don't set any codes. Um
Yeah, you'll have to reach out to your EM- your EMR CSM, so whoever supports you from the, um, CS perspective or even sending that question into our support email, so support@prompthealth.com. Um, they'll [00:58:30] be able to help you understand that 'cause I'm not quite sure exactly if you'll be able to do that or not, um, like preset them.
Um, yeah, so send that in. If you have a moment, do it right now while it's on your brain, um, 'cause then you'll get, you'll get a quick answer on that. It's a good question.
.png)
.png)
.png)