S3E5 Proactive Patient Outreach with AI (ft. Steve Smith, INOVA)

Jordan Cooper   0:53
Thanks. So I understand that you're using this AI digital assistant to build out a new phone system, and I'd love to hear the story about kind of where you were. Who decided that there was what the problem was, and then how you came upon AI digital assistant as the solution and then we can get into how that's been implemented.

Smith, Steve   1:15
So, you know, we've heard loud and clear from our patients that they're trying to access our system in a, you know more highly efficient way and you know what you know in the addition to doing you know various scheduling, open scheduling, ticket scheduling, mychart scheduling which is kind. Of the standard for we really had a need to provide patients more self-service and more autonomy when they call us and be able to provide them options to self-service, whether it's scheduling cancellation. Rescheduling. Make medication refills in a more automated way. And so last.December, we went live with a new phone system.Nicest the phone system and it's powered by Jiro, which is the digital assistant the other. Work case was that we did had a tremendous amount of just transfer calls, right where patients get transferred from department to department and that was really becoming troublesome. So the intent again is to allow the patient to interact with the digital assistant. Let them speak their responses and get them routed to the correct appropriate destination without kind of going from department to department, and so that's that's been the impetus for this.

Jordan Cooper   2:26
So it sounds like there were two main KP is for success, one was.More efficient delivery of services. And responsiveness to whatever the patient was trying to do. Be it scheduling, rescheduling, medication refill, and the second was a diminishment of the burden on your team and trying to pick that up with the AI digital system, is that right?

Smith, Steve   2:52
Exactly right, exactly. Reduce calls into the contact center and be able to allow patients to navigate through and journey more efficiently with the digital assistant.

Jordan Cooper   3:01
Got it. So.

Smith, Steve   3:02
And I would just add that we roll out in December and.

Jordan Cooper   3:05
Mm hmm.

Smith, Steve   3:07
We've seen immediate gains, whether that's successfully routed calls where the patient interacts with the digital assistant states their intention, and we're able to route those calls efficiently.Upwards of 80% across the organization. So you know we're exceeding expectation in terms of efficiency and getting calls routed to the appropriate destination based on what's spoken in the event that the digital assistant can't connect with the patient, it's still defaults to our call center. So you know, they're still going to get serviced. It just won't get that automation. There's just a backstop with the contact center.

Jordan Cooper   3:45
Got it. So do you have any metrics on what the average number of transfers were before you implemented this versus now or what the length of the average call was before versus now or any other sort of specific indicators of these gains?

Smith, Steve   4:04
Well, I will tell you that when we compared January 24 to January 25 S January 24 with the old phone system to January 25 with the new phone system, we were able to reduce about 50,000 phone calls across the board.And also to your point, we were average handle time went from about 7 or 8 minutes to about 5 or about 6 minutes. A little under 6 minutes. In the same time we we had more scheduling encounters year over year from from. So with the new phone system less calls but yet more scheduling encounters. So, you know, really a win win across the board in terms of transfer rate. That's that's gone. We've reduced that by about 9% so far. We're still working through. We're not fully baked in with the goal. Why? But from what we do have with the digital assistant, we currently also have. Cancellation. Patients being able to cancel their appointment, confirm their appointment and reschedule their appointment. We just don't have some additional add-ons, but those will be coming here in the very short term.

Jordan Cooper   5:17
Mm hmm. So some of our listeners to healthy data podcasts will be wondering, all right. You know, we also have trouble with patients having to be transferred too many times and our call centers are overwhelmed. And this sounds kind of familiar is resonating with me. But I've looked at solutions in the past and they're expensive or difficult to implement. The rollout is Rocky. There's some hiccups. Can you talk about the time to implement or how if you did it in house or your referred it out, if there were any problems with rollout or everything's just smooth? And they should just. Our listeners should just jump in and go right ahead with trying to replicate your success.

Smith, Steve   5:58
Well, I would say go. Lives are never smooth and that that you know, that's just part of. Nature of the beast. But what I will say is we spend a significant time with an RFP. We vetted many platforms, they had adhere to our security requirements as well as integration within EPIC. Our you know our EMR and so those were kind of the qualifications. In terms of advice. We, you know, certainly having dedicated resources from the vendor I think is paramount having clear. Timelines and and ramifications of those timelines aren't being met, and having a really strong project manager. Is is essential, you know, as I said, it wasn't perfect.

Jordan Cooper   6:51
Was this less than a year? Would you say was the implementation less than a year or six months or a year and a half? What was it?

Smith, Steve   6:58
I'd I'd say about. I'd say upwards of a year.

Jordan Cooper   7:02
OK. And then once you, I I I know that since we're on the topic of AI and then we'll we'll switch to the other use case in a moment. But on this topic of Gen. AI.

Smith, Steve   7:10
Sure.

Jordan Cooper   7:13
There's been a lot of discussion around the country with ambient listening and integrating voice data directly into the EHR.I'm wondering if your AI digital assistant in any way enables extraction of anything communicated by a patient over the phone through voice. And if that can be converted into discrete data elements in any place, as sometimes happens with a chatbot or is that still kind of a future milestone?

Smith, Steve   7:42
So. We're we're about to roll out next week. As a matter of fact, a workforce management. Which will have, excuse me, ambient listening and we will be able to extract what's said and what's not said. So for example, you know we have a KPI and we really focus on demographic validation within the scheduling process. And so there's certain things that we would expect our agents to ask within the construct of of the call.

Jordan Cooper   8:06
Mm H.

Smith, Steve   8:12
And we'll be able to pull calls and identify individuals that aren't seeing the the appropriate or omitting, you know, part of the demographic validation. So we're going to be able to really drill down on who's saying it and who's not. And then provide get the calls and have them listen to that. So QA process developments, but also it's important for us to start the scheduling process. Off on the right foot. And so that's important. Part of demographic validation. I'm really excited about that ambient listening and the and analytic coming down the Pike.

Jordan Cooper   8:48
Now I think there's something else. Something else that's interesting that you're doing with your AI digital assistant, which I think you referred to as Hero. You said that it's able to provide translation services to patients and you said that providing better access to patients in the way they need care has been a prior. Tell me about that service.

Smith, Steve   9:05
So the digital assistant is named Hiro Hyro and we are in the process of working through with the platform meta being able to attach.

Jordan Cooper   9:08
Mm hmm.

Smith, Steve   9:21
That within the construct of the of the phone call with with Nice so and it's integrated through HIRO. So you know, right now our translation services a bit clunky where patient needs translation services. Because our agent needs to put them on hold. Call third party vendor you know create the conference call. Sometimes those calls drop course with a human. You don't always you know you don't have, like a standard, you know. Of performance, right? Whereas some translators might be really good, some translators might not be as good as good. So we are in the process of building out with meta. AI component to the phone call where if a patient needs you know Spanish or what we have like. 50 or 60 different languages that they can. They'll speak, they'll make their declaration, and they need assistance with translation.

Jordan Cooper   10:06
Mm hmm.

Smith, Steve   10:11
The the the bot will turn on and then it will just be a really a seamless phone call with the scheduler where we can take the transferring out of the equation. I think that would also certainly reduce average handle time with with the call as well. So I'm really excited to see that get rolled out probably towards the end of the year is what we would expect.

Jordan Cooper   10:32
That was excellent. I'm wondering you're you're trying to make Innova more responsive. The patient needs to make patient experiences better. I guess that would improve your HEDIS scores. Are there any other financial, clinical or operational drivers to improve the patient experience with their encounter with the Nova Health system?

Smith, Steve   10:55
Well, we have a AKPI of 12% online scheduling and you know little by little over the course of the last few years we've we've been improving in the in this past year, we've we've eclipsed the 12% online schedule and I think that's in. Large part because. Of all the the changes we've made, whether you know through it's through ticket scheduling to my chart.

Jordan Cooper   11:27
Mm hmm.

Smith, Steve   11:27
We're gonna actually really blow that number out of the water here later this year when we roll in. Organic scheduling or ticket scheduling within the context of the phone system where the phone system, the AI digital system, will recognize who's gonna call who's calling in based on the phone number, and if there's an open referral in the system, the AI will see that or and and. Prompt the patient to say, hey, I see you have a referral on file. Would you like to go ahead and schedule so? So that I think is going to be really great. You know, our patients have told us that they want more access and we've worked really, really hard to do that. And I think you know the proof is in pudding because recently we were recognized as the press G.

Jordan Cooper   12:00
Mm hmm.

Smith, Steve   12:12
Hospital of the year in large part is because of our surveys with patient access and how we're doing. So that's, I mean, it's really exciting to see that all the hard work is actually paid off. Got it.

Jordan Cooper   12:23
Excellent. I appreciate that. So you also said I think I wanna, we'll come back to workforce management, but on this topic of improving patient access, let's talk about epic. Cheers for a moment. So you're implementing this epics new CRM customer relationship management platform in order to be more proactive in a patient journey? Document phone calls and interactions with the patient.And I think that's all kind of a means of achieving the end, which we've been discussing. Which is improving patient access and patient satisfaction. Can you tell me about the impact of implementing cheers and how that's been received by patients and providers?

Smith, Steve   13:06
I think it's been received really well. So it's really kind of cool because the agents actually answer the phone call within the cheers dashboard. So they don't have to go like from like platform to platform. But actually, if the phone nice phone system is integrated within the cheers call hub, they're able screen pops come up. So we actually can see when the patient's calling and who is calling in so. It take you already know who you're talking to, and so it really kind of expedites the validation. Process because this that that screen pops and then every phone call is if it's a, you know, clinical or a financial matter, we would expect our team members to create acrm or document the phone call. So the next agent that may answer a call from that same patient will be able to kind of see the timeline additionally. We're able to provide patients what we call next best actions and so if something's like needs to be. Updated within the patient account whether like their their phone number is maybe a cell phone or their e-mail needs to be updated or their insurance is expired.

Jordan Cooper   14:14
Mm hmm.

Smith, Steve   14:17
That populates in the call hub. So our agents are prompted to ask the patient, hey, do you want to update your insurance? I see it's out of date. Or hey, once a year. We ask patients about their communication preferences and how they'd like to be communicated, whether it's phone text, e-mail, all the above, and so the next best actions prompts. Our agents, to be more proactive within the context of the phone call and then ultimately making the experience for the patient more seamless and possibly even reducing follow up phone calls for them to call back and update the account or provide or get guidance on how to do. So.

Jordan Cooper   14:51
So, I mean I'm hearing that. Cheers is useful for improving data integrity. You said it's expediting the validation process. I heard there's a lot of business reasons that cheers is providing value to Innova like prompting the the patient to update their insurance information or to contact information. I'm wondering if there's any clinical benefits to cheers. For example, if Epic fires a best practice advisory to provide.That this person's. A. 1C was elevated and maybe they need to make an appointment to see if their feet are swollen. I need to come in for the diabetes management or if there's other chronic disease management is is cheers being used to drive clinical outreach to say hey, this is a potentially high risk patient. You know you're you're over 50 years old, let's get you a colorectal screening maybe. Screen for cancer.What are the clinical implications for cheers, if any?

Smith, Steve   15:48
Well, if for example, if a patient hasn't scheduled their annual visit, that will populate with the next best actions to prompt the agent to ask and get that scheduled. I would also say that we're able to communicate with providers and the office through a clinical encounter, so we.

Jordan Cooper   15:57
Mm hmm.

Smith, Steve   16:09
Able to actually create the encounter within shears. Send that to the provider, to your point to provide feedback. Get guidance. Of a patient needs to follow up. With the provider, we're able to generate messages from the cheers platform directly to the provider or the provider's assistant.

Jordan Cooper   16:27
Got it. And so our call agents, the only ones using the cheers platform or our case managers also using it.

Smith, Steve   16:34
Right now we've just rolled it out to our call center agents. We have talked about rowing it out further within the organization, but that's to be determined.

Jordan Cooper   16:43
Got it. OK. And has it been helpful in acquiring and retaining patients and preventing patients from leaking to competitors in the region?

Smith, Steve   16:52
It has. We're actually about to roll out campaigns within shears where we can send out based on their communication preferences follow up. So I don't orders that need to be fulfilled. If they have a follow up appointment, they need to schedule if for. Patients in need or they have transportation services need needs. We we we. For folks that might be at risk and we can provide services for that. So really excited to to do that and to your point, I. We I don't have a broad data in terms of how many patients we've acquired, but I would say is once we start utilizing the campaigns feature, we will be able to see data in terms of how successful the campaign was.

Jordan Cooper   17:42
Mm hmm.

Smith, Steve   17:45
So that's going to be coming later this year and I think that's going to really be a a big, big.

Jordan Cooper   17:46
Got it.

Smith, Steve   17:52
Impact.

Jordan Cooper   17:53
I'd like to transition to the final topic that we're gonna cover in this podcast episode. You did allude to it earlier, and of course I'm talking about your recent go live with a workforce management platform. Or actually, sorry, you're going live with it next week in mid-May.

Smith, Steve   18:06
Yeah.

Jordan Cooper   18:09
There are a few goals that you had associated with that you're looking to forecast staffing. There's a few other goals you mentioned. Could you talk about the implications of, I guess the motivations? Let's start with of why you implemented this new work. Management tool and what you're hoping to achieve.

Smith, Steve   18:25
Sure. So first, just day-to-day intra management of the team to see to see exactly where team members are doing according to our KP. Is that all that data will be populated within the manager dashboard and as well as the agents will see how they're performing so.

Jordan Cooper   18:40
Mm hmm.

Smith, Steve   18:43
If they're expected to have an 80% availability rate, they will see exactly where they are. We have a real focus on.

Jordan Cooper   18:49
Mm hmm.

Smith, Steve   18:52
Eliminating calls under one minute because. Without workforce management, you know it's hard to be proactive in terms of folks that you know are not adhering to our processes. So we'll be able to populate both within the agent dashboard and within the manager dashboard calls under a minute and and agents agents know that they're that that that is a real focus on our organization, so.

Jordan Cooper   19:18
Sorry, just to drill into that for a second. If a patient calls and the call lasts less than a minute, does that mean it was a mistake or that they were dropped or does it mean that this could have been taken care of on my chart, what what?

Smith, Steve   19:20
Yeah.

Jordan Cooper   19:32
The thought process there.

Smith, Steve   19:34
I'd say a couple things. One, it could be a mistake. Two, was it someone quite frankly, trying to avoid work and game the system when we don't have tools like workforce management to proactively manage the team in real time, you know in any call center, there's always folks that are looking to you know not answer calls or and. So that and we've identified that as an area that we need to focus on a little bit more deeply. So now.

Jordan Cooper   20:03
Hmm.

Smith, Steve   20:03
With this new workforce management, we'll be able to see who has, for example, the most calls with you know in within the day of one minute or less. Then we can go in and start drilling down and listening to those calls. And was it a technology issue for the agent? Were they? Or were they actually trying to avoid calls? And we'll be able to root that out very quickly.

Jordan Cooper   20:24
Got it. So it's accountability in the call center.

Smith, Steve   20:26
Yeah, you got it.

Jordan Cooper   20:27
There could be call center agents who are intentionally dropping calls I see.

Smith, Steve   20:30
Right, right. Yeah. We, we again every call center. There's always folks that are kind of trying to game the system and workforce management is gonna allow us to be more proactive and get on top of that in real time.

Jordan Cooper   20:36
Hello.

Smith, Steve   20:44
And so there's no way, quite frankly, to hide. And we what we want is people to answer phone calls and work with our patients to provide them great service. The other thing in terms of workforce management is to be more proactive in terms of how we prepare for the day in the week based on historical call. Data and we're going to be able to implement our payer platform, Kronos within WFM. So we'll be able to see from historic data based on also who's like taking PTO. What the week or the day ahead is going to look like? And then that will give the managers power to say, hey, it looks like next Wednesday I'm gonna be a little short staff based on the fact that folks are on PTO. It's gonna be a high number in volume. I'm going to need to maybe use some resources outside of my, you know my group.

Jordan Cooper   21:25
Mm hmm.

Smith, Steve   21:31
So we have a real focus on cross training and be able to use that in real time and that's where the the scheduling component and forecasting will come into play.

Jordan Cooper   21:31
Got it. So Steve, we are approaching the end of the podcast episode and there are a few more things I want to get in under our belt before we conclude. You mentioned ambient listening. Usually when most people are thinking about ambient listening, you're thinking about a cell phone app in the clinical encounter with a provider standing there, and instead of typing on the computer, the app is kind of documenting a lot of the progress note for the. Patients directly into EPIC. But here you're talking about ambient listening as a Gen. AI solution for a call center. Can you talk about that use case? Cuz it's a little bit unique.

Smith, Steve   22:12
Yeah. So this is gonna be really, really cool, so. We'll be able to track. The context of the call like based on segmentation. So for example, the system will recognize, for example, if the calls become combative and there's friction, and the manager will be able to see that in real time and then they could potentially start listening in the call because they're seeing an alert that there's some friction. On the call, they can actually jump into the call based on that, but will be able be more proactive on deescalating phone calls and right now and and that's also going to hold our. Again, our team members accountable because you know we want to work with patients and. We, you know, every agent is in is, you know is, you know, doesn't work as well with patients is is the next one. So that getting that information in real time when we can say that we can see based on what's being said that the conversation. Kind of going off the rails a little bit.It will prompt the supervisor manager to listen, possibly jump in and then from QA development standpoint we can pull all the calls that have those you know, elevated metrics where that's occurring and again provide coaching development opportunities.

Jordan Cooper   23:31
Huh. And so it sounds like from a management perspective, there's a lot of value to be had here with with keeping call center agents accountable.Is there any pushback from the call center agents themselves who say this isn't gonna help me? This is gonna help my boss kind of manage me. Are they seeing value here too?

Smith, Steve   23:51
Well, here's how we approach that. Is that? We're in the business of answering phone calls and providing patients care and reducing, reducing pain and suffering. And if you are not on board on those values.

Jordan Cooper   24:09 Mm hmm.

Smith, Steve   24:10
And that's that. That's going to be a concern. So we're we know that we most of our agents are doing really well and doing the right thing. But what it but but the why and what we're telling our agents is. It's we're going to find the folks that are bringing the team down or the few bad apples, so to speak, that there's no place to hide and we're here to provide patients a service.

Jordan Cooper   24:26
Mm hmm.

Smith, Steve   24:31
And if you're not here for that, with all due respect, I'm not sure you're a good fit on our team. So you know that that's, you know again it's been well received, most of our agents do a phenomenal job and are here for the right reason. It's rooting out the folks that are trying to game the system. And bringing this, you know, bringing more work to the folks that are doing the right thing.

Jordan Cooper   24:53
So Steve, we are approaching the end of the episode right now. We've covered a lot of ground. We talked about the AI digital assistant decreasing transfers we talked about. Routing to the right position automatically taking care of Med refills and scheduling. We discussed AI digital assistant helping with translation, again reducing hold time, dropped calls, transfers. We talked about workforce management in order to again improve the patient experience. When they do call in. And we also discussed the cheers CRM which is again to be proactive. The patient all of this is a very patient centric approach. I think that's what's motivating all the topics we discuss. Can you leave our listeners with a parting thought about if they would like to make their organization more patient centric and improve their patient satisfaction scores? What message would you have for them?

Smith, Steve   25:46
That. The the healthcare industry is obviously behind in in terms of like the Amazons or the airline industry in terms of, you know, being proactive with with you know with technology quite frankly and but that's, but that's the norm now that patients customers want to handle things within their. Phone don't necessarily need to want to speak to a live person. And it's it's it's an incumbent. Oh, I also health care leaders to meet patients where they're at and give them the ability, provide them technology to access our system and access our system at a high level. Because if we don't, we have potential leakage and they go find another health system that does have that technology and provides them that service.

Jordan Cooper   26:35
Thank you. I appreciate that for our listeners, this has been Steve Smith, the AVP of enterprise contact center and access at Innova Healthcare delivery system. Steve, I'd like to thank you so much for joining us today.

Jordan Cooper stopped transcription