We're here today with Julie Smith, the corporate IT director

of clinical applications and support about this health care

for those who don't know about this.

Healthcare is headquartered in Pensacola, FL and comprises 3

hospitals, four medical parks, a complex network of primary care

and specialty clinic practices, and an Institute for Orthopedics

and Sports Medicine.

We have an interesting topic to discuss today.

We're going to be discussing alleviating clinician burden

with the creation of help desk desk tickets that leverages Gen

AI technology.

So Julie, thanks for joining us.

And please, let's dive right in.

Tell me a little bit about Cassie.

Cassie.

Cassie is our newest team member.

I always ask people if we can introduce them to Cassie.

Cassie stands for the clinical application support interactive

bot, so she is she is called Cassie by those of us who know

her well.

And tell me about the genesis of Cassidy.

What was going on about this health care prior to the

introduction of this Gen AI tool?

And what were some of the problems that (Baptist

healthcare was facing?

So we had when we implemented our current EHR, we had

contracted for a third party EHR support vendor for help desk

specifically for their product.

There's.

And as we were watching the Slas and getting the reports every

month that comes in, what I noticed is that I was seeing

whole times of two and three minutes and you know Max wait

times and and then how long it was taking to get some of those

tickets even opened, right.

And we all know that our clinicians, their focus is not

on calling the help desk.

It's not on trying to resolve a problem that they're having,

it's all about the patient care and they have more work than

they usually need.

It.

And so I started thinking about how can we take that off of

E.

their plate?

I've been looking at reduction of burden.

Overall, I think most healthcare systems have, but that was one

that I really wanted to tackle and take away.

So that was really where it all started.

Got it.

And just so our listeners are clear, these help desk tickets

are with clinicians requesting support within their electronic

health record, which is epic.

So there's some kind of issue that is either requiring

immediate assistance for patient care or they can't close out an

issue and they'd be waiting on the phone with the patient in

the room.

UM for sometimes you said 3 minutes or somebody to get on

the phone and just listen to what the issue is and it would

take even longer to resolve these issues, correct?

Brought.

I'll correct one thing.

We're not on epic.

We are on Altera sunrise throughout our enterprise, but

Terrorist.

otherwise yes and and they may may or may not be with a patient

Turner.

when they're making those calls.

Sometimes it's an education question.

They just need to know how to get past a step that they're

working on, and sometimes they might get an error.

So it does kind of just stop their workflow and it's tracks.

Right.

OK, so you have an issue and then you know your team comes

around and says, well, there's there's, uh, you, you partner

with Microsoft 365 and you have you've learned about copied

studio and you have some ideas about how Gen AI could augment

the internal help desk and move away from a vendor help desk.

Can you talk about the workflow with the vendor help desk and

how you propose to make this a budget neutral request to

leadership so that you may might be amenable to this suggestion.

Would be happy to also mention the fact that you at first our

first thought wasn't.

Ohh well we have Microsoft and we can use to copilot studio.

Umm.

That would have been so much faster and easier if we had

gotten to the solution that quickly, but really it was

trying to find out.

Asking ourselves, having conversations.

What is the fastest way and the fastest way?

Was they just need to give you a couple of pieces of information.

How can you get a hold of me and what the problem is?

And then just keep moving and that's where the thought of the

Umm.

bot came on.

As you mentioned, we wanted to do, we needed to do a been to a

budget neutral approach.

I mean, we all see that in healthcare and and have for

several years.

How critical it is to manage that budget?

So I wanted to start by looking at our current vendors and did

they have something that they could help us with as you

mentioned, of course we have Microsoft 365 and so we took a

look at what they had to offer and then we also had to

understand well if our our providers are still calling a

third party service desk outside of our walls then is the chat

Yeah.

bot going to do as much good as we'd hoped and we came to the

solution that it's not we wanna bring that clinical service desk

actually back in house.

And we can talk a little bit about that as well.

That's in the complement to Cassie.

But then we took a look at how can we build this copilot studio

bot?

And we went to the vendor vendors are supposed to be our

partners.

How can we do what we need to do in the most efficient manner and

we were looking for somebody who could partner with us and help

Well.

build the bot because we wanted to roll this out and in full

from the beginning, making sure that we skipped any hiccups when

we rolled this out to our clinicians.

Got it.

And so you were able to develop Cassie, this clinical

application support interactive bot.

Tell me right now, how is Cassie interacting with your internal

help desk right now from the perspective of a clinician.

So can you walk me through an actual storyline of I'm a

clinician, I'm in the patient room.

There's some kind of issue and my patient sitting right here.

What do I do?

We we have within Citrix right now, but soon and I say soon

then the next week we'll actually have a launch point

directly within their EHR.

So that it is within their workflow that they can select

this to launch Cassie.

It already knows who they are.

So because you know, we work with Microsoft already, it says,

Perfect.

hey, we know that this is who you are.

This is how we can contact you.

This is your email address.

And then it just gives them simple options.

The options are do you want to open up a ticket?

We all know physicians are busy, so we've also offered do you

Umm.

wanna open up a ticket on someone else's behalf?

So that office managers charge nurses, anybody who might be

working with them, maybe some of their support staff could go

ahead and do that for them as well.

So they could open up a ticket or they could search for

Take a look.

additional information.

All they do.

OK, so let's say I have a problem.

I'm trying to prescribe a controlled substance.

We have to go through ECS for that.

Uh-huh.

I'm getting an error and so I need to open up a ticket.

The nice thing is that caffeine knows who you are, so she

already knows that you're a physician, and so it actually

gives me an option.

Do you need to open up a ticket?

Yes.

Is this for general clinical information or is this for ECS?

It's for ECS.

Give me your contact information.

At that point, they may actually put in their office manager's

name or number.

That's fine.

We understand again, they don't want to be on there long and

then they just let us know in a sentence or two what the problem

is.

So two clicks my contact information and a sentence, and

within 2030 seconds they're back in their regular patient flow.

Got it.

Or they could even stick their head out of the the patient exam

room and say, hey, you know, office manager, nurse, whoever

happens to be around, would you mind going into Cassie and

opening a ticket for prescribing controlled substance and then

they could do it on a clinician's behalf?

Absolutely.

It's just of 10 second request to a colleague and they can go

ahead and do it even though they don't have the prescription

authority.

They can open the request on behalf of the requesting

provider.

That's right.

Now, if it's a general, how do I use the system?

Maybe they're newer.

They haven't used control substance prescribing very long.

Perfect.

Umm.

Cassie does have the ability to search through all of our

education documents.

We have hundreds of education documents, so if she says if the

doctor says, oh, I just need a little bit more information,

that's great.

What do you need information about?

He or she types in ECS and it says here's all the information

we have on ECS.

What can I help you with?

So self serve I don't have to call somebody and wait for a

call back.

So many of our listeners listening to this episode maybe

wondering I love this new technological solution.

It seems to expedite things.

It seems to solve a real problem.

It's budget neutral because it's replacing an external vendor

call center that you've now gotten rid of and scaled back,

and you brought some more stuff in House leadership likes it.

They've approved it because it is budget neutral, but the next

question that a listener may come up with is can we talk

about change management and provider adoption?

What is provider responses bin like?

Anecdotally, what sort of feedback have you received?

I know you've done a soft roll out, but maybe promotion is

still on the horizon.

How have you been managing kind of adoption of Cassie and have

there been any pushback at all saying, hey, we're used this

other workflow, how are you, you're changing it on us?

Has there been any any difficulties or challenges with

implementing Cassie?

That's a great question.

We took the approach of of really a marketing approach.

Umm.

One of the reasons why we gave her the name Cassie, we've also

for adoption used in AI avatar vendor third party.

Umm.

We just subscribed to it.

We're we're able to put in some some scripts, so it's engaging.

So when we have announcements over what's coming out new in

the EHR, if we do have maybe a scheduled downtime, any kind of

plans we give that to them in an engaging way that has helped.

Again, they're busy.

Sometimes they don't want to listen to even a one or two

minute video.

Overall, they've accepted it because it's a if they love it,

use it.

If you don't go ahead and call the help desk, you still have

that workflow option or turn to the next person beside you.

So from a physician standpoint, we haven't seen it big uptick in

adoption, but I'll say our soft rollout was about a month ago.

So we're very, very new into this where we have seen a great

Umm.

deal of adoption is in our physician clinics.

So the support team that they work with, as you mentioned the

So.

doctor leaning out and saying, hey, I'm having a problem call

OK.

the help desk.

Umm.

Those are the team members that we're seeing a lot of it.

And today, actually this week we are taking the time to round on

our nursing units.

Our nurses are busy.

They don't see emails, you know, they're worried about their

patients.

And so we're going out to them and when we round, we also

Umm.

mention it too.

My told my team, oftentimes we'll get a phone call as a

OK.

clinical application support and sometimes we say, well, have you

opened a ticket yet?

I told him from now on what I wanna hear if they say ohh I

don't have a ticket.

Say, did you tell Cassie cause Cassie will open up the ticket

for you?

Right. So.

There's two things I want to follow kind of tie up this case

study here.

One is, I understand that there are some metrics for measuring

cassies effectiveness that may be presented at HIMSS in the

future.

Do you?

Can you tell me something about what you're thinking about and

how you will evaluate the effectiveness of Cassie and how

your leadership will be evaluating certain KPI that

you're collecting pertaining to this, this intervention?

Yeah.

Thanks for that, Jordan.

We did just submit a proposal for HIMS 25 and you know next

Umm.

year down in Las Vegas and what we're hoping to be able to give

is a very robust amount of KPIs around the use of it.

We already know that it is saving minutes every time a

physician needs to open up a ticket because we specifically

tracked when we had one of our providers opening up tickets.

We wanted to know specifically what was their wait times.

Everybody's wait time is important, but of course we

focused on those.

We know that the wait time has significantly decreased.

We're continuing to do studies on that.

Uh-huh.

Corrently, when they send in the information, they tell Cassie

It.

what the problem is.

The.

Cassie opens up a ticket.

Initially it's going into a queue to be reviewed, but very

Umm.

soon our plan is that will directly go to those teams, so

it's going to directly go to the clinical team.

It's not just clinicians using this, it might be anyone who's

in patient access registration or in other areas so they can

pick where this ticket needs to go and it doesn't have to stop

and a queue it can actually get to the person and get assigned

quicker.

So those are some of the things that we're gonna be looking at

is how we can expedite the help to everybody who's using Cassie.

So in addition to measuring the effectiveness of Cassie, as I

mentioned earlier, you know with the, with the kind of change

management efforts, I understand that you spearheaded efforts as

a primary leader for a carousel of change for EHR optimization.

That has included collaboratively working with

physicians specialties to customize the platform for their

specific workflows.

Can you tell me about the origins of this carousel of

change and how you've used it in other projects and then how you

brought it to bear in the cache project?

Yeah, the carousel of change has been around at Baptist for

several years and it's because there is such a clamor for

optimization of the EHR.

Everybody wants what they're using optimized and So what we

wanted to be able to do is to be able to focus on specialties.

Umm, you know anyone who works in the EHR realm knows that you

can't customize every document for every provider or every

order set.

So what we've done is we've taken that carousel, the

specialty gets to come on board, talk to us about what are the

pain points, maybe what's been underutilized, what they're

spending the most time on and then we're able to optimize

Umm.

those and sprints.

And once we have that optimized, then they get off the carousel.

Uh-huh.

We let the next specialty on the carousel, it's their turn to go

around and get optimized and then we come back around.

And were you able to leverage the carousel of change at all

with Cassie?

Or is Cassie the product of the Carousel of change?

What was there any interplay between the two?

Cassie is really more cross the entire organization, but and so

Umm.

I wouldn't say it's necessarily connected to the carousel of

change.

It's really connected to that decreasing clinician burden, but

Umm.

you mentioned change control.

So I do wanna go back to something you mentioned earlier.

We haven't touched on which is generative AI, and I think that

Umm.

when it comes to AI, you know, anybody in an IT in healthcare

IT is talking about AI in some way.

Of course you need to have governance.

We do have a a steering committee when it comes to AI

with Cassie, we were able to take the generative AI.

Of course, we spoke with compliance, we spoke with our

security team.

We made sure that there wasn't any concerns and what she was

able to generate and what we've done is really given our

organization that first step into generative AI, but with a

very restricted amount.

So we provide her the documents that she can read and she can

generate information for us.

She can't go out and search the Internet and let you know what

the weather is in Orlando right now, but she can, however, tell

you what the directory is.

Do you need help for biomech support?

All you have to do is ask her what's the number for BioMed?

She'll let you know what that is.

Got it.

So it's she has a limited pool of data upon which the machine

learning algorithms are trained that are defined particularly

specifically by the governance, compliance and security teams.

That's right.

And that's the first step.

That's where we are right now.

You know, I think that just talking in general, not

necessarily specific to Cassie or our health system, you get to

stair step that in based off of governance based off of how

Problem.

comfortable people are.

So you might want to start with like.

We are very distinct library of information and what we do is we

upload we have two or three documents.

Anytime we have new information that needs to go out through

those documents, we take that through change control, just

like we should.

And it's great because the generative AI, we can upload

That's.

information.

When is our next upgrade?

What are we doing for hospital week this week?

Umm.

Even questions such as that general information can be

uploaded to her, but I could see the next step being something

that might be a resource for our clinicians that they already use

and letting it look at that website and I I'm kind of

hesitant to to go out on a limb and say any vendors cause

stitely not why I'm here.

But.

But you know, we personally use umm, you know, a vendor for

education and and clinical knowledge being able to maybe

the next step.

Take a look at can we allow Cassie to use that website with

peer reviewed information on it?

The.

That might be the next step, but we're not there yet.

So if you were to just looked in the future towards what could be

done with Gen AI app, this healthcare beyond help desk

tickets, it sounds like you might be looking for educational

purposes to pull in marketing information.

Look at peer reviewed academic medical journals.

I have this particular patient and this specialty practice and

I heard that there may be a study here, but is there any

other clinical trial that my patient may be eligible for?

Do you know?

Are there any kind of thoughts towards a road map of an

interoperability road map to pull in different kinds of data

sets and give that Cassie access to different kinds of data that

either that may not live within the EHR?

Yeah, Jordan, I I hope you will come back and ask me that a year

from now because we, you know, we really are so new to rolling

Cassie out.

I think she has been extremely successful and those that have

started using her use her everyday.

Umm I I think where we can go is I don't wanna say limitless, but

possibly I think all of those examples you gave are perfect

examples of where I would love to be able to get to umm and

then you know on and and and this isn't specific to (Baptist

Umm.

I I don't want it to make make it sound like it is.

But you know what I've heard?

And I love this as I'm rounding in the hospital.

I have people saying, oh, this is great.

Can we get a bot or can we just put our information in Cassie?

Hey, can we get a bot that we can put out HR information and

Umm.

so people are asking me questions and I love that.

It's making them think about it in a way they never have before.

So as we approach the end of this podcast episode and you're

talking about how other people are reacting to Cassie, I'd like

you to speak to our listeners for one minute and just what are

some of the lessons learned or what do you think is one thing

they should be aware of as they consider bringing Gen AI to

their clinicians and their own organizations?

What advice do you have?

Oh gosh, anybody who knows me knows I can never just have one

point right? Umm.

I would say make sure that you are in contact with your

security team.

Your IT compliance.

Make sure that there aren't questions that are abounding

that maybe get ahead of where you are and stay on top of that,

but also in encourage those questions and so we are also

Yeah.

partnered with a vendor and one of the things I looked for in

that vendor is I wanted them to teach us.

I didn't want to vendor that.

I needed to attached to within, uh gosh, they need to be here

and they need to support us forever.

1st.

I wanted them to say, hey, let us teach you how to do this so

The.

that you can support yourself and you can build and you can

expand on your copilots.

And so that was something that I would encourage as well.

The more you know is, the more you know.

Sounds like a pathway to independence vendor who teaches

you how to fish instead of giving you a fish.

Exactly right.

Alright.

Well, thank you very much.

Uh, for a listeners, this has been Julie Smith, the corporate

IT director of clinical applications and support at

Baptist Health Care.

Julie, thank you so much for joining us today.

Thanks Jordan.