AI Agents Reduce Patient Access Time and Pajama Time for Doctors

Rohit: Hi Crystal, great to see you. Thank you for joining us. As you might be aware, Crystal, the show—is called The Big Unlock. It was started by Paddy Padmanabhan of Damo Consulting, and we are building on the foundation that he had laid. And it is great to have you here. I’m Rohit Mahajan, I’m the Managing Partner and CEO at Damo Consulting and BigRio. Would you like to please introduce yourself to the audience?

Crystal: Hi Rohit. Thanks for having me on the show. Hi everybody. I am Crystal Broj. I’m the Chief Digital Transformation Officer at the Medical University of South Carolina. I am based in Charleston, South Carolina, but our system spread the entire state—all 46 counties of South Carolina. We are one of the oldest medical universities in the country, and we’re celebrating our 200th year this year. So, we’ve been around a long time. We have 2.4 million patient encounters annually, 16 hospitals, and about 2,700 licensed beds. But besides being a health system, we’re also one of two national telehealth centres in the country, recognized as Centres of Excellence.

We have a university with six colleges, about 884 residents and fellows, and 42-degree programs. We’re the number one rated hospital in South Carolina according to U.S. News & World Report. We’re also a research organization with about $300 million in research funding annually, 1,200 clinical trials, and about 55 active startups. There are a lot of exciting things happening here, and I’m just so blessed to be here, bringing digital transformation to the organization in a variety of ways. 

Q: Can you please tell us what inspired you and attracted you to this space? How did you get started in healthcare? Tell us a little bit about your journey, where you’re at, and what you might be thinking about for the future.

Crystal: I started out as a COBOL programmer when I came out of school. I was in IT, doing COBOL and programming. Then I jumped around to a few different jobs—back in the day, if you didn’t move, you’d get stuck in the system. So, I moved around, went into consulting, then back to big business, then back into consulting. I did that for a while, but with two young children at home, working 60 hours a week in consulting wasn’t sustainable. So, I took some time off and had the opportunity to be a Director of Christian Education at a church for a couple of years. It was totally different—a passion project.

I don’t have the typical IT trajectory. I bounced around quite a bit. Eventually, I went back to management roles, working across various industries—manufacturing, health clubs, insurance, human resources companies. So, I gained a broad range of experience. I didn’t grow up in healthcare. My first step into the healthcare space was as Chief Technology Officer for the American Association of Diabetes Educators. It was healthcare-adjacent—we trained diabetes educators to help patients. That was my first real jump into healthcare.

Then COVID happened, and like many, I found myself out of work for a while. During that time, I actually wrote children’s books.

Yeah, that’s the fun fact—I saved it for now. I put them on Amazon. I probably made a hundred dollars, but it was fun. I wrote and illustrated them, and it gave me something creative to focus on. Afterward, I ran a business analyst office, and we started teaching Agile. Then I went to NorthShore University HealthSystem in Chicago, where I was Vice President for Digital. I built a digital team from scratch—there wasn’t one before—and we grew it to eight people, supported by some consulting services.

Then my dream job came up in Charleston. I wasn’t actively looking, but my daughter was attending school here—she’s earning her doctorate in Nurse Anaesthesiology and graduates next month. We started spending more time here and fell in love with Charleston. When this job opportunity appeared, I knew it was exactly what I wanted.

I started as a team of one. This organization is very innovative—they truly believe in innovation. What we’ve done in just two and a half years is incredible. My team has grown from one to eight, and I’m lucky to work with amazing people—not just my staff, but also the senior leadership who constantly ask, “What can we do next?”

The team that works for me puts their whole heart into everything they do, and it shows. When we roll out projects, they’re rolled out carefully and thoughtfully. One of the things I always say is: We won’t roll it out until it’s right, because we only get one chance to make a good first impression. We don’t just throw something out there and hope it works. We started with a small project. We were, we were using Notable is the, is the key name, right? You, you’ve heard of them, obviously. They’re well known in the marketplace.  We started with a pre-engagement process. So, if you have an appointment, we send you a note three days before— “Hey, here’s a reminder. Do you want to confirm or cancel your appointment?”

We launched this with just five offices to see how it would go. We wanted people to try it, to prepare the offices, and to understand how patients would react. It began as a pilot, and then we started to grow from there. One of the biggest lessons learned was: start small, then scale. From five offices, we moved to 100, then 500, and now it’s implemented across all offices in the organization. We send out pre-reminders, and we receive confirmations and cancellations in advance. Then we added another layer: validating demographic information. We pulled data from Epic, so patients could quickly review— “Yes, I still live at the same address,” “Yes, I’m still married,”—all the things providers need to know before you walk in the door.

After that, we integrated bill pay. Patients could pay their copay or any outstanding balance in advance. So, you can see, we added small features step-by-step. We didn’t try to perfect everything at once, but each step was rolled out thoughtfully and intentionally.

We also listened to patient feedback. At the end of each interaction, patients could give a thumbs up or thumbs down. And so patients would say, thumbs up, and then they’d say, this was great, or this is this. We get some thumbs down and we read every single one from our customers. We make sure that we’re touching the patients and find out what they want. And so we got things like, Hey, I come to the doctor three times a week. Why are you sending me the same information all at once?

So, we worked with the vendor and adjusted it. Now, at the beginning of the week, patients receive a single message listing all appointments: “You have three appointments this week. Do you want to check in for all of them now?” They can fill out all their information at once and then simply walk in and say, “Hi, I’m Crystal. I’m here.” We also added Spanish support. Spanish-speaking patients now receive the messages in Spanish, and the feedback has been fantastic—things like, “This is easy,” “Thank you,” “I can do this from home.” It’s been a huge success.

But one key thing we learned was the importance of change management. Our front desk staff were used to handing out clipboards. So when we rolled out digital check-in, some staff were unsure—“How do I know it really worked?” Some still wanted to give patients the clipboard. That, in turn, frustrated patients who had already checked in online. They’d say, “Wait a minute—I already did this!” So we had to step back and think through how the technology impacts everyone, not just the patients. We retrained front desk staff to trust the system and support the new process.

Of course, there were a few glitches—nothing is ever perfect. Sometimes the system didn’t work, and staff would say, “Just come to the front desk,” which undermines the experience. But we’ve been learning and improving every step of the way. It’s okay. Sometimes we’d get someone new at the front desk who had no idea what this was. They hadn’t been part of the rollout, so they didn’t know, they didn’t understand it, and it just wasn’t on their radar. So, we created training materials for new hires. Now, whenever someone joins, they receive the necessary training. We also make sure to communicate updates—like, “By the way, patients can now get this in Spanish,” or “Here’s a new feature we’ve added”—so that everyone is on the same page. Those were some really valuable lessons for us: start small, communicate—and over-communicate—and grow alongside the vendor to create something truly impactful.

Q: That’s great to know Crystal. Lot of great experience there. In terms of scaling the solution, I did read somewhere that you mentioned that innovation is in in our DNA. Could you talk any other of you Digital transformation projects or innovation projects where you’ve had success and you, I think you are very good at tracking metrics as well, is what I understood. So how, how do you track those metrics and kind of look at the return on investment?

Crystal: Sure. I’m happy to go into a bunch of those things. Once we started with Notable doing that, then we started looking at how we could do other outreaches to patients. In Covid, people didn’t go to the doctor, right? So, women especially don’t take care of themselves. We went back to close the care gap for mammograms because most women didn’t go during Covid, and then they got busy and forgot about it. So, we did outreach. If you haven’t had a mammogram in a year, we send you a reminder saying, “Hey, we care about your health. You’re overdue for a mammogram. Would you like to schedule it now?” Automatically schedule it. You didn’t have to call in. You didn’t have to do anything. We turned it on at five o’clock at night on a Thursday, and by the next morning, we had 129 women that had scheduled their mammograms. That’s a KPI. And then we started tracking those women to see. It’s still in progress today. We do it monthly and go, “Who hasn’t gotten it in a year?” Every month. We send out these notifications. Up to date, I think we’ve had 18,000 mammograms that have been scheduled. What we found is that about 180 of those women needed to come back because they had abnormal results. So, we got them the care that they needed before it became too late. That’s where you’re like, okay, this is digital. And there might be KPIs, but these are KPIs where we could potentially be saving people’s lives. And there’s no dollar amount that you can put on that, right?

And so, with that being very successful, we started hitting up well-child to make sure people came in for their visits, for their kids getting the shots they need before school. We send notifications about flu shots to remind people to come in and schedule. We’re going to start doing uncontrolled hypertension and getting readings on that so people can do it—diabetes as well—start checking on that. We’re just continuing to build on that platform and get those kinds of things.We know that we have sent out, since June last year, 1.7 million reminders to our patients. We’ve had about 50,000 cancellations. That seems like a lot of canceled appointments. But at the same time, if you cancel, it’s not a no show. The doctor can refill it. We have an automated way to do that through Epic. It’s called Fast Pass. You can get somebody else in. That helps with access because then people can get in sooner. Helps doctors’ schedules. Makes everything flow better for both patients and providers, which is really what transformation should do, right? We don’t ever want to burden providers. We want to make things easier every month. Easier for patients as well. With Notable, I can’t say enough about it. We have a 98% satisfaction rate from our patients, and that’s another KPI that we track. We just started with a new product with Notable where we’re actually working with revenue cycle—we’re doing prior authorizations.

We all know that you might get a referral for an MRI, but you have to wait for your insurance company to say it’s okay before you get it and make sure they’ll pay for it, all that kind of stuff. Well, to do that, somebody has to take your referral, type it in Epic, and then go to the payer site and type in all the information again. And play back and forth with the payer, then put it back into Epic. That can take anywhere from 15 to 30 minutes. But we have an automated AI agent that does that now, and they can do it in about 30 seconds.

If you think about how much faster that is for the back of the house—not just that number alone—but how much quicker patients can get in. And tracking how many are automatically approved right away—we have about a 37% accuracy on this agent, and it keeps learning all the time. That means almost 40% of the ones we send through are done and the human doesn’t have to touch it.

Coming up next, we’ll go, “Okay, Rohit, you’ve been approved. Here’s a link. Schedule.” We’re going next with that. So those are some things that we’re doing. Great success with that really helped our senior staff know that this is a team that can take big things on and do it well. We brought a voice bot into our patient access center. We have 42 phone lines, about 150 people that answer phones every day. If you ever call the access center and you press one for this, press two for that, and wait on hold for who knows how long—we’ve got an automated natural language processing agent.

We call her Emily. She answers the phone: “Hi, I’m Emily, your digital assistant. How can I help you?” You say, “I’m checking on my appointment.” She says, “Okay, give me some information to validate who you are.” It goes into Epic. “Yes, Crystal, I see that you have an appointment next Tuesday and next Friday. Which one do you want to know about?” “The one with Dr. E.” “Okay, sure. That’s two o’clock next Tuesday. Do you want to confirm, cancel it, or do you need to reschedule?” “Actually, I just needed to know where to park.” She can do all of that, and it doesn’t take up an agent’s time.

We deflect—and that’s another thing we track—we deflect about 17% of the calls that come into the patient access center. We’re not getting rid of jobs, not by any means, but it means that our patient access reps can handle more complex questions. Something that’s harder to do. Our hold times have gone down. The number of people that get frustrated and just hang up has gone down. So those are the types of things that we track to prove that the software—these digital tools—are doing what we really want them to do.

Q. That’s interesting implementation. So, the handoff between Emily and the real live agent is seamless?

Crystal: Yep. The thing about Emily is she can answer the phones 24/7. So if the center was only open, I don’t know, eight to five or whatever. So, she can answer after hours and answer questions and maybe sometimes it’s a more complex thing.

So the agent can go like – sorry, I can’t, you know, have you call back to talk to an agent tomorrow to say you call kind of thing. And we’re just now changing it so that she can reschedule because      rescheduling is much harder than just scheduling. And, within the next month she’ll speak Spanish too.  But that comes with another thing where we really had to train the bot. Right. So, it’s, it had to learn. It must learn. Accents. We’re in the south. I come from Chicago, so I don’t sound like I’m Southern except when I say y’all. But we had to train it so it understands the accents. It would understand certain words, understand that appointment may mean visit may mean something else. And do that training of the model. And take the time to do that. So, testing is important. Yes. Really going through and testing and validating that to make sure that it works. Getting the people that are going to use the software to use it, so they understand what patients are hearing so that when it comes to me, I’ll know that, oh, Emily already asked this and this, and I’ll have some of that information for me.

Things like that. But it’s wildly successful. All 42 lines on it. Now we’re going to roll it out to the rest of the state because Right. Oh great. Just in Charleston, we’re going to roll it out for our revenue cycle. We’re going to roll it out for our pharmacy.  So that we can take some work off the back of the house. So again, they can work at a higher level of license, whatever that license is. So that growth stuff that that things that we just want to get out stupid stuff and then start to do, you know, those higher level things that provide more value to both our system as well as our patients. 

Q. So, I heard you say, I think early on in our conversation, maybe even before the podcast started, that you, you did look at a lot of KPIs and you, you have a certain way of looking at return on investment and there’s certain benchmarks that you kind of track. So, could you tell us a little bit more about that piece?

Crystal: Some of it is based on how many patients encounters we’re facilitating or the financial impact we’re seeing. For example, with pre-experience copays, we’ve collected about $1.4 million through the technology—completely automated, no human involvement.

Front desk staff often don’t like asking patients for money, but if you collect ahead of time or send a reminder afterward, patients can just use their credit card—done. For open balances, we’ve collected about $1.9 million. For voice, we track how many calls are deflected, how many are completed, and how much labor time we’re saving. A phone call can take three to five minutes. If an employee doesn’t have to handle that, they can be more productive elsewhere.

We also track bot performance—how many calls it picks up, how many referrals and authorizations it handles. We compile all of this into a monthly report for our stakeholders, including business leaders and senior staff, showing how many appointments were scheduled, how many tasks were completed, and the overall impact of the technology. It clearly shows the value and why we continue investing in it.

We just launched new scheduling software last week. If you visit our website, uchealth.org, and click on “Find a Doctor,” you’ll see the new interface powered by a product called DAX Care. You can search by specialty—primary care, ortho, etc.—or use natural language, like “I have an elbow strain,” and it’ll pull up orthopaedics with first available appointments, both for new and returning patients. It pulls directly from Epic and displays real-time availability, making appointment access completely transparent.

In just one week, with no advertising, we’ve had over 200 appointments scheduled through the new system. People are finding it on their own, and satisfaction is high. We started with primary care and are rolling out specialties next—orthopaedics tomorrow, then new specialties every couple of weeks.

We expect strong results from this rollout because patients everywhere are asking for easier access. Now, they can see exactly when a doctor is available. And if someone searches something like “I have a cold,” we surface not just primary care, but also options for virtual care. So if they want to be seen now, they can click and do a virtual visit—something many forgot about after COVID.

Virtual care helps alleviate scheduling pressure since we don’t have enough doctors to meet demand in-person. If your child gets sick at 7 p.m., this gives you quick access and guidance on what to do next. Speaking from experience as a busy mom, having that option is incredibly valuable.

Q: That’s awesome. So, what are some of the challenges, crystal, like you go about implementing all these solutions? What are some of the big challenges that are being faced by, you know, people who are implementing these solutions?

Crystal: I think, across the country, you know, we’re short staffed. Our IT departments are short-staffed, so you know, they’re busy keeping the lights on and they do an amazing job making sure epic, workday and all those things are working and going plus. All the technology from the laptop I have to the network and all that kind of stuff they must handle as well.

So, you know, they’re doing that. And then they’re busy full-time. It’s not like they’re sitting around. Right. And then here comes digital transformation and they have a new toy that they want to have. Right. And they want to put it in. And it’s like, it’s not like I can go do that on my own. Like I do all the work around it but eventually tie into a main system. So, it’s, you know, sharing with. It why this is important, what we’re trying to do, and getting on their schedule so that they can connect me to a Cadence analyst or an API analyst who you know, get cooked up together. But it’s not like the projects they’ve done in the past. You know, if you do an upgrade for a major system, you know, on June 30th, I’m going to, you know, implement the system.

So, 30 days before, I should be done with testing and 30 days before I should do something else. When you start using these new technologies, like I said, you start small. Five offices. So okay, I need a connection for the five offices, then I’m going to go in in two months to a hundred offices. Well then I need a new report that gives me a hundred offices, and then six weeks later I need another report added on for these many things. So, I’m like the gift that keeps on giving to IT. Which does not always make me popular. Exactly. And yet, you know, you know, we found a way to kind of work together and we’re still working out some of the fine tuning on that to make it really sing because it’s hard for them to figure out what does digital transformation means now, right? An API person, you know, the vendors will tell me, oh, this is easy. It’s four hours, you’ll be done. Yeah. Four hours to get the connection going. And then, you know what? When it doesn’t work, I’ve got to give you a call so you can talk to my analyst to go, why does troubleshoot. That’s it. They can’t plan for that.

Right. And so, we don’t know. Well, we might have problems, but you know, they say they won’t, but we probably will. So, let’s see if we can figure out, let you know that this might happen. But there’s no way I can predict what the problem’s going to be or how many hours I’m going to need. And that’s, that makes it, I think a lot of people are struggling with that across the country. And that whole agile concept of, we’re going to add more. It’s sort of, it’s never done. It makes it hard to budget resources. You know, my team’s just constantly churning through it, but it’s got like a whole other job to keep the lights on. That’s a whole different. So that’s a little bit of a challenge.

And then working with our user side, our business side. You know, we’re changing the way they do things. And change is hard. And so, there’s a lot of change management. Why are we doing it? What’s in it for me? Why should I do this? And some people resist that. We’re using ambient technology as well here. Not with a pilot of 125 doctors that are using, nuanced stacks and they’re talking in their phones and saying, okay, Rohit, I’m the doctor. You’re my patient. So this is going to record my notes. So, if you’re okay with that, and we always ask. And so, press play and then you and I talk and like we did before, we were talking about my daughter’s graduation, we’re talking about other stuff.

And then I’m going to ask you about, I don’t know, maybe you have asthma or something else. I’m going to talk to you about that. I’m going to talk about the medications, how it’s working, how it’s not, and then tell your dog I said hi, tell whatever. And all that gets translated into the note. Except the stuff about my kid, about your vacation, about your dog isn’t there? It just takes the medical stuff. Which is still amazing to me that AI can do that. And then, you know, you can say yes no, or adjust stuff. There’s always human that are just checking the ai cause it’s a consistent, it’s not, you know, an independent body. Right. And then you pick that and it’s that; well it saved 37% of pajama time for doctors.     

Yeah. Which is crazy good. Right? And it also closes charts faster. So, revenue cycle’s really happy about that. When we did our pilot, not every doctor loved it. Yeah. You’re like, uh, and of course they don’t want to look dumb in front of their patients. Yeah. So, if they aren’t comfortable with it, they’re not going to use it. And so, there are a lot of things that we learned from that, that you know, okay, we have to find out what the sweet spot is. How do we communicate? We wrote great emails, the most beautiful emails in the world, Dr. Crystal here, this is what you need to know, and here’s some information. And we’d send them out.

The doctors don’t read it. They don’t. So you, they weren’t responding, they weren’t asking questions, they just stopped using the software. So, we had to figure out ways to either text or have somebody in their office ask them how it’s going. And some doctors just didn’t take it right away, so then we gave it to somebody who wanted it. Overall, the amount of time savings and closing charts faster and doing those kinds of things to give doctor satisfaction. At the end of the year was great. It was so great that we’re giving an enterprise license. We have 600 doctors waiting to get the technology. Which we’re going to start rolling out this spring. So, the KPIs were important to show that this really was valid in the pilot to, there was a no go, and if it wasn’t working, we weren’t going to use it. We showed that it did work and so then we can add on to something else.

Q: That’s great initiatives, Crystal. So, as we are coming now to kind of close off the podcast, I would like to ask you for, you know, what do you see when you look into the future? What is your peek into the future? And any other closing remarks or thoughts that you would like to share with the audience? Crystal?

Crystal: I think we’re going to see more and more AI, but I don’t think we’ll ever see AI taking over our jobs. I think we’re going to have to find some sort of a sweet spot where integration becomes easier between our main systems and these add-on systems. I think go away from like, if you go to any show, you see, oh, there’s this for, I don’t know, brain cancer. Let’s say there’s this for heart, there’s this for whatever. And you know what? I can’t afford to buy this. Nor can I integrate those. Do that kind of stuff. So we have to have something that can kind of do everything at least somewhat well, but we can customize and make it so that it’s better for the whole system. I think we’ll see more of that coming along. The spot solutions, not that they’re not great ideas, but I can’t afford to do it. I don’t know many hospital systems that can take on that many things.

I think we’ll see more in patient navigation, and I don’t mean wayfinding per se, but I mean like I know I’m going to have to go in for an ortho appointment, so you’re in education before that, I’ll be able to maybe answer my questions through two-way text or something beforehand. I’ll get stuff a day before or three days before. Maybe my husband who’s going to support me will get something as well, and then I’ll come in, maybe there’s something in the hospital that does some education or whatever, and then when I’m discharged, I’ll get follow-up both from a nurse. Maybe it’s text first, and then I have a question. A nurse will call me back. Maybe I’ll get some more education. Then all of that will be seamless to me as the patient. It may be pointing solutions along the way, but for me it’s just seamless. It’s my hospital system caring about me. It’s my hospital system showing me the way to go. Those kinds of things we’ll see a lot more of that coming out as we get better at using the tools that are at our disposal.

Subscribe to our podcast series at www.thebigunlock.com and write us at info@thebigunlock.com   

Disclaimer: This Q&A has been derived from the podcast transcript and has been edited for readability and clarity.