PegaWorld iNspire 2023: UnitedHealthcare – Proactively Reducing Healthcare Fraud

The National Health Care Anti-Fraud Association indicates that over $300 billion annually, or 9% of all healthcare spend in the United States, may be fraudulent. Learn how UnitedHealthcare got proactive at tackling this problem – with people, process, and technology. Hear how their transition to Pega helped the organization improve efficiency in its operations team to apply more intelligence and automation for even greater impact.

Transcript:

- Well, this is the last session of the day and I know from my conversations in the Innovation Hub, this has actually been heavily anticipated. So I thank you all for joining us. Again, welcome to PegaWorld. I'm really looking forward to this presentation. Couple of housekeeping items. So first of all, we have three incredible speakers. This story from UnitedHealthcare is phenomenal. Kristen is actually joining you from vacation.

- [Kristen] Yep.

- She's coming in to speak with you. Housekeeping item. So again, this is the last session of today, I know we're keeping you from the bar, which opens at four, so we'll have that. Questions, if you would hold those to the end and also there are mics up front. Again, this is videoed and the questions as well. The sound is taped. It's really important and valuable for folks to hear the questions. So if you do have questions at the end, and again, I'd like to have you hold those to the end, but please use one of the microphones here up front as we go to the the Q&A session. Without further ado, it is my distinct pleasure to introduce and welcome David, Brent and Kristen from UnitedHealthcare. David.

- Thank you. Thank you. Oh, good afternoon everybody. So we get to come and share a little bit about our program that we have at UnitedHealthcare called Enhanced Provider Validation. This is a core fraud, waste and abuse program within UnitedHealthcare. And one that we really believe is kind of at the tip of the spear of where we're going in healthcare fraud today. And it's a great partnership that we have with Pega in helping us kind of achieve sort of our mission here. So when we jump in for a baseline, just a little bit about healthcare fraud in general, it's estimated that about 9% of all healthcare dollars go out the door are fraudulent. So doing anything we can do to fend that off becomes critical work. In 2018, we set up this program called EPV. It was tied out of a fraud scheme that came up through the DME section within UnitedHealthcare. We could have caught it very quickly had we had the right protections in place, but we didn't. And out of it came this need to say, why are we not doing a better job at knowing who we're doing business with? We're really good at knowing the type of business we do, but are we really that good at knowing who we do that business with? And so, the program got stood up in 2018 very quickly, thanks in large part to Brent Cooley here, and it's already achieved close to half a billion dollars in overall savings over the five years. And one thing that we'll kind of come back to at the end of the presentation as well is at the front of this 79% of all the providers we began identifying were durable medical equipment suppliers are DME. So we'll kind of hint back to that here at the end. So, EPV. So right now this is our process for assessing new non-participating tax identification numbers for fraud, waste and abuse upon the submission of their first claim to UnitedHealthcare. So within the healthcare fraud arena, we've really got this goal of trying to be as proactive as possible or how do we get to essentially the point where we can just avoid it happening altogether. So generally speaking, when we develop healthcare fraud cases, we start off in the post pay place. So all the way to the right there. Post-pay investigations, you usually have to have a lot of claims data, a lot of paid claims to start identifying a Barr billing patterns. Once you identify those Barr billing patterns we go after and we try to recoup 'em, we make sense on the dollar doing that, it doesn't do a whole lot. And once we catch 'em then we're like well haha, let's stop it going forward. We'll put them on prepay investigation. That's where we're gonna stop all the claims. We're not gonna pay you until we can get your medical records in the door and make sure that you are documenting what it is that you are billing us for. In that scenario, the fraudsters are really quick on this. They know how to get around this. There's so much AI talk around it today. One of the things we're getting out in front of right now are physicians who are using ChatGPT to simply write out medical records to circumvent coding reviews and documentation audits. But the gold standard of where we really wanna get and where we talk about shifting left is avoidance investigation. And that's really EPV's focus is that shift left. And to get there, it started by really focusing on how we engage with our providers. And we took a step back and we said, when we receive a claim that's no different than you going into the bank and presenting a cheque. And when you go to the bank and present a cheque to the bank, they're gonna want an ID, they're gonna wanna make sure that you are who you say you are. If you're opening an account, you've gotta produce a social security card, you've gotta produce a birth certificate, all of these things that validate you. We don't do that on the healthcare space very well. So we started from that perspective, what can we be doing to validate who those physicians are that are coming to our front door? And what we learned was a byproduct of that is we really start to find the bad actors, the fraudulent providers. So our program is set up in these core areas here. I'll go through these very, very briefly, but what we focus on too is the circle here, the holistic approach, that white line, our white circle is really our technology. This is how we deploy not only our analytics but our workflow management tool within Pega to really aid in what we're doing. So we start very upfront. We have a method of identifying all new tax identification numbers that come into our company. Once we identify it as an unknown provider to us, it moves into our workflow management system where now we've got a team of folks that are going to see that tin all the way through. That authentication team that's down there, they will review roughly 57,000 new tins a year. And out of that, we can apply different sorts of risk models to who that provider is. Not so much what they're doing but who they are. We look at very basic things. Does your tax ID match with what the IRS says? Do you have a license? Do you have an appropriate MPI? These are all things that we can build into the model and help us score out the risk of having this provider with us. Those providers that reach a very higher level of risk, we can do this through rules engines, we can do this through all sorts of analytics scoring. We send them on for further investigation. So this is a second layer investigation where we're sending it actually out to our investigative staff. We call them the AIU team or Avoidance Investigative Unit. And this group gives us the opportunity to go put boots on the ground, actually go visit some places, determine is this a lab? Is this a DME? We actually wanna see. And that level of investigation is pretty much is critical as you can get to our process because the outcome of an investigation that goes to our AIU team, goes to what we call FFC or this is a Fraud Flag Committee. And here, we're gonna potentially make the determination. Do we wanna hard deny this provider? Hard deny for us means we don't wanna do business with you. You submit a claim to our system, it does not pass go, clearly does not collect $200, you will not have a relationship with us in this organization. So the scrutiny for us to reach that level of a hard deny on these avoidance investigations is very high. The success rate has been astronomical as well. We were just looking at some numbers the other day. Our AIU team accounts for 50% of all the hard denies that happen within UnitedHealthcare. The other 50% is all that stuff that's happening to the right. That's the post-pay investigations, the pre-pay investigation. So our EPV program is meeting in some cases exceeding the hard deny rate of our seasoned investigative groups within UnitedHealthcare. So I'm gonna turn it over to Kristen and she'll talk a little bit more about how we've integrated with Pega.

- All right, thank you Dave. Good afternoon everyone. My name is Kristen Nolan, and I'm a manager within the payment integrity area. I've been supporting now the EPV team for about three and a half, four years. Currently, I have a product owner role within the team to work with our Pega development team here within United and and Optum. So what I wanted to talk about today, it's all about Pega. We evolved to Pega. We didn't start with Pega when we started this program. Like many new programs just getting off the ground, we started tracking cases on just Excel spreadsheets. Basic information, passing it back and forth. Everybody's been there, I'm sure in this room. We quickly evolved because it wasn't working for us to using a SharePoint list so at least multiple people can work on the same information at the same time. And as you'll see too, we have multiple teams in different layers working on these cases. And that's really where Pega comes in. So really our third generation is when we built the Pega application we have today. It is an application workflow tool. So we can easily move our cases once they're created within the program. Between what Dave said, the authentication team, the investigation team. We had an onsite team as well for when it was justified that we actually had to go physically on location to do an investigation. And then of course back to the investigator to the FFC if that warranted it. So it was a lot of back and forth things and that's where Pega really helped us in moving those cases back and forth. And then we are quickly going to be entering our fourth generation and a lot of going with Pega and we're doing a lot more enhancements and bringing AI to the application and moving our analytics upfront so that we can work with more data than we do today. So it's super exciting, especially in this time right now. Lots of work to be done this year. This is the design of Pega. So we have six different stages. I kind of briefly talked about this, that you know, we start with a manual intake. We can get a case through a download that we get from our claims highway, we call them. We can manually create a case, or we also have a portal that I'll talk about in just a moment. So there's three different ways that a case can be created. From there, it moves into authentication. From authentication, it can go into investigations or it could go and be implemented. We could move into that implementation stage. It can move back from investigations to authentication, or onto onsite, or investigations to mitigation. So there's a lot of ins and outs here where the case can go. Ultimately, it's going to be implemented. But that's how the design of Pega first started. That's our groundwork. I wanted to talk quickly about some enhancements that we brought to the application over the last probably two and a half years. These have been really focused on how can we bring... You guys... Thank you. Can you hear me? Did it mute? It's okay. Okay, great. So with the enhancements, these are just three that I picked. We have been deploying, last year alone we did 27 deployments. We've already done seven or eight already year some small, some are bigger. But I pulled out three that we've done in recent months that were extremely beneficial to the program to bring efficiencies, or just improve the user's experience. That's also very important to me as a user of the application myself. We wanna make sure our users, our authenticators, our investigators are being able to do their job easily. We don't wanna hold them up by something not working correctly. So in 2022, January, 2022, right outta the gates, we implemented the Cosmos user interface. So we updated the look and feel, so it was easier for our users to use the application. Was very well received. Later, we implemented the associated cases. This was huge for our program because up until then, we had no way to link our cases within our application together if we had multiple tins with the same NPI number, or a phone number, or an address, we have learned and have many examples of fraudsters that create a new identity but will use their same cell phone, or the same address or something like that. So this was really huge for us to put this in and be able to make those ties and put that very easily out in front of our users. So our investigators can easily just click on a button, click on the button and they can see all of the other cases that have similar values. Right now we're building on seven values, but in the future, we will have more than that. So that's been very helpful and it will help us as we expand and grow to be able to handle more data to bring in more association. Redesign screens. We had sent a survey out to our users asking, what are their pain points within the Pega application? And we came up with three, three huge pain points for them. Two of 'em were around claw logs. Just tracking their call logs to the providers or the members, patient interviews, member interviews. So we redesigned those screens and rolled them out with the Cosmos user interface. And now, coming up the end of this month, we have a huge deployment where all of the research that's captured within the case, all of the investigation work will be redesigned to instead of a freeform text, we're actually getting standard fields. So now we can report on them and we can grow our associated cases feature. So that's super, super exciting to us. We've also done a lot in the automation. Automation is big for us and it has been a focus for us for last year and this year. We created an out-of-network portal. In 2021, we created a space for our out of network providers to actually go put the information that we were asking of them. So we sent out an RFI letter, which we automated late last year asking for all the information that we want. You're a new non part in coming in. We have things that we want, a W9 and some various other things. And so that is automated now where the authenticator is not manually trying to work the case and then send that letter out. So that saves a lot of time and efficiency within the authentication space specifically. That ties to the outof network portal. So the outof network portal is where the provider can go. They put their information in there and through our integration with the Pega application, it gets fed right into Pega, which is fantastic. We're not looking at faxes, trying to manually input things into the case anymore. It's all coming directly in through Kafka. And then notifications. We are really leveraging this now so that we can eliminate email traffic and instant messages between the users when we are routing cases back and forth to each other between the teams. So that is a huge thing and we're continuing to explore that and leverage on as we are putting in automation, we are always thinking about how we can expand that notification. So, future state. We have a lot of exciting things that are gonna be kicking off this year. Huge things. And so we have put together what we think the future state, what I talk about this fourth generation could look like for our program. And the yellow dots here, you can see is kind of what we're changing. So we want to improve the demographic risk model and we're bringing in behavioral risk model, and growing that, moving it upfront so that based on that information that we're getting and we have large amounts of data coming in, our workflow tool can route those cases around based on the information and also prioritize our cases. So we're working on the right cases at the right time. That's gonna be huge for our program moving forward as we grow. And then with this as well, we're looking to integrate our authentication and investigation teams together so that we can have that cohesive end-to-end type of case review there. So with that I'll hand it over to Brent, to talk about our successes.

- [Brent] Yep, thank you. Thank you Kristen.

- [Kristen] Yep. So my name's Brent Cooley, I'm one half of the team of six simul black belts that took this from vague idea to the program we've been learning about today. I'm also the current program manager. And now that Dave and Kristen have done all the hard work of preparing for the party, I get to waltz in with a beer and pizza so we can talk about successes. So yeah. So what we have here is a chart. I told you guys I was a black belt. This is how we party. So, I know this is a boring numbers slide before anyone walks out falls asleep. Thank you for saying not at all. I wanna assure you this is the only boring numbers slide here. In fact, this is the most boring piece of the most boring numbers slide. But I'm starting with this because we are in Vegas, talking about successes, and there's not a single dollar sign on this slide. And the reason for that is, money is not always an indicator of success. Now, I'm not saying that because I want to be your life guru, or because I'm trying to console you for your performance at the tables last night, Dave. I'm saying this because specifically for fraud, we have to take what the environment gives us, right? We can look at past history of savings, we can math that forward to some forecasts on what we think might happen this year. But we cannot set goals, targets or quotas for savings. The reason is, who here has been pulled over by the police gotten a speeding ticket and you know for sure that's because they were trying to meet their end of month quota? Anybody? Yeah. Yeah. Yeah. You know how that felt? Terrible. Multiply that feeling by a Fortune five company and that's how it feels if we are in court and a jury sees that we set savings quotas. It does not matter what the bad actor was doing, you can throw the merits of that case out, their siding against us, right? So, can't set targets. In addition, the better we do, if we're really on target, the better we do, the worse we look. Remember that now. So we're coming back to these numbers. We have to find success some other way that's not money, right? And so we have numbers like this, we're gonna go back to the beginning. What we did here, was we stood this up in a very short period of time. And as a black belt, we want to take an infuriating amount of time looking at data before we make a decision and pull the trigger on something. But when we are in a space that has never been done before in the company, and absolutely not been done before at the scope and scale, we did it within the industry, there is no data. So as a black belt, if you're trying to make decisions on little or no data, what you need to be able to do, is you need to be able to see in near real time what happened after you made that decision so that you can say, I told you it'd be the right idea, right? And with Pega, we were able to get that data near real time, we make a decision, we got a result. We make a decision, this is super easy. We make a decision. Oops! So this is a non-black belt decision right there. This is a success. We needed to move fast, we needed to make calls, we had little to no data when we were doing it. We needed to know when we messed up so that we can back things out as quickly as possible. Learn those lessons, learn from our oopses and try again. One way of measuring success. We also need to move fast, right? We talked about moving fast. We had to get this stood up. I think we had the minimum viable product, that stage one that you were talking about in three months. It took six months after that to get what you would call a pilot type phase, which is a stage three. And we kind of soft launched from there. Speed was necessary. Why? Here's a press release from the Department of Justice about a company that took Medicare for multiple millions of dollars. This article was July 6th, 2022. Because of the speed that we were able to work, we got hit for nowhere near millions of dollars, orders of magnitude less, right? Pega was the solution that was able to deliver the capabilities we wanted at the timeframes and speed we needed so that we could get in front of this. And this isn't a one-off, it continues today, it is consistent and repeatable. CMS alert, here you go from April, 2023, alerting us of Elevate, ProCare and PharMax. It's really nice whenever you get an alert like this and you realize that we're probably the ones that tipped them off on what was going on. Now, we also need to be able to manage large amounts of data. Data coming in from the providers, data that we need to be able to share left and right. Here's an example of what we can do managing that data. There's a provider in Alabama and we are having a little bit of trouble with getting their address nailed down in Google. So we ask them, can you just send us a picture of your storefront? Right? And this is what we received. Can you read that? I cannot. But one of our investigators, I know he pulled some real CSI stuff man, enhance, enhance. And he was able to find on Google Maps the picture of that storefront. If you look at the pattern of the lights, the pattern of the reflection in the windows, the first picture is a cropped version of the second picture. Except this guy was in Georgia. The Alabama provider did not actually exist. All because we could pass data back and forth seamlessly from our folks, our authenticators that were taking the data in, to our investigators that were looking at it, from the provider that was trying to defraud us. Now this is great. This is the totally legitimate drop shipping warehouse for a totally legitimate, durable medical equipment company. We've been talking up to this point about durable medical equipment, alright? And in case you're aware, here's what the scheme is. Someone will call your mom or your grandma, and they will ask her if she needs a knee brace. And if you're lucky they'll call. Sometimes they'll just send it to her. She will say, "No, I feel fine. I don't need a knee brace." And they will say, "Great, we'll send it right out." And she'll say, "No, I don't think you understand. I don't need a knee brace." And they'll say, "We're from Medicare, we're from UnitedHealthcare, we're from whatever. It's free. And if you don't need it now, you will. We'll just send it out." And they'll send her the need brace and or the back brace, the leg brace, the neck brace. In fact, there's a term in the industry called a mummy kit for when they get every brace. So, why did we focus on DME first? Why is this such an issue? This right here, UHC number one, this was on a post-it note on the cubicle wall of one of these call center reps. UHC number one. We love being number one. We don't love being their number one, right? And so we focused on durable medical equipment and here's how that's gone. Now, the red line going down, that's a good thing. We love that. But what this trend kind of hides a little bit is that this right here in 2019, that was our baby steps in Florida. All of those stops were in Florida. And here projected for 2023, is national. Every provider nationally. Now, we're pretty sure we haven't gotten worse at finding durable medical equipment providers, right? We're pretty good at that. It's where we started, it's our bread and butter. But, if you remember what we said before, the better you do, if you're really hitting it in a targeted manner, the better you do, the worse you look. There have been some environmental factors, a couple Department of Justice take downs, things like that. But we don't believe that accounts for every bit of this decline you see in our hard denies. This is a deterrence effect. This is those bad actors taking their UHC number one foam finger and quietly putting it under their desk, right? So this is able to be done. Again, we are able to get, pick it up fast. We're able to share documents, we're able to manage our workflow effectively to target from the thousands of tax ID numbers we get every year, target specific provider types in specific regions, in specific areas, with specific criteria to say we need to find the bad actors. And over time, the better we've done, the worse we've looked. Money in the case of a healthcare fraud is not necessarily an indicator of success. Now, if we pull this all together, it's a nice looking house. I kind of sucker for hardwood floors, although the candles do get a little bit of grandma energy, but that's not a lab. How many people here shop at Forever 21? No one. Oh there's, yeah, there we go. Have you ever found a chiropractor in there? Yeah, neither did we. Yeah. Both of these facilities, these providers, looked legitimate at first glance. They weren't sending people every brace ever, right? And calling attention to themselves. They didn't look that bad at first. It would've taken a long time for our analytics to find a pattern of billing that was untoward, and that's if it did. But with our ability to validate the providers and do something as simple as, I don't think there's a chiropractor in Forever 21, I don't even think they have it in the back with the shirts and the size you can't find it. This is what we're able to do with the help and the assistance of the capabilities of Pega. And this is all in the beginning, right? We have in the works, new techniques, new AI, new capabilities that we're taking to bring this to a much larger, more effective and more efficient place. So, we have plenty of time for questions. Alright, we went through that kind of fast. So, does anyone have any questions about anything we've talked about today? Don't be shy. Hi.

- This is the first time I've seen any presentation related to fraud. So I applaud you guys for what you've done. Absolutely brilliant, right? I mean I've seen, I think it was on TV, this show about it was an investigative report about fraud and how, you know, companies in Florida set up involved, you know, doing this kind of thing, especially the DMEs. So you talked about these fraudsters using different ways to cheat. I mean, now with ChatGPT and AI and stuff like that, have you encountered anything that's coming through that path, where they're starting to use different ways to cheat or fraud?

- Well, number one, we know they like to use tried and true methods. And the worse they do at it, the better we like 'em, right? So what we've seen are kind of evolutions of existing methods. We were expecting maybe a little more ChatGPT stuff than we've seen so far. Not to say it's not out there, but a lot of times you're looking at forged W9s, you know, with that DME scheme, with the call center. They call people and harass them and everything. Just so the person will say, "If I say yes, will you leave me alone?" And now they're covered because the member ordered it. That doesn't really work. But, you know. So short answer, not big yet. We're still seeing evolutions on known stuff. But they do surprise us at times. So we'll see.

- One other question. So your final outcome, is that blacklisting the provider, or does it go beyond that? Like going to the local authorities or federal agent?

- It can certainly go down that path. So within any investigation we do, we have a regulatory requirement to do necessary reporting. So in a lot of those examples that Brent was sharing, that's where we think a lot of our reporting upfront was what, you know, could have aided in those investigations and getting closed out. The end result too is at the end of the day, you know, we want to have a good provider network. So at the end of the day, I want the best provider data that we have to know that then we have the best providers we've got out there too. So, you know, I wouldn't use the term blacklist. I mean the hard deny is our biggest outcome where we just decide not to do business with you. We have an administrative outcome as well. You saw on the first slide where we separate out the authentication process and the investigation process. Within the authentication process, if we have providers who fail to validate with us, they'll be rejected. Which simply means that their tin is not allowed in our system. Their claims are not allowed to flow onto our platforms. That accounts for about 22% of the volume that we receive. And again, we look at 57,000 tins a year. So, you know, that 22%, that's a deterrence factor. Sometimes it's non-par providers who are just like, "Eh, I don't care, it's a $30 claim." So there are multiple outcomes we're always trying to go for, but at the end of the day, I'm always looking to try and just get the best provider data we can, validate it, authenticate it, be able to apply that fraud risk to it and know once they're in there, we've kind of done our job and now it's up to our other areas to continue to monitor. We refer to ourselves as a screen door. And I'm from Minnesota and screen doors are big in the summer 'cause of the mosquitoes. So our goal is to just tighten that screen down as best we possibly can. And maybe one day we get to a steel door, I don't know. But right now we're a pretty good screen door.

- Thank you.

- Thank you.

- My question, how did you deal with people stealing real data. There is credentialed websites where you can scrub data and you can get the real information. So did you guys use like bank accounts? How did you figure out they were fake? If they have real tins, real NPI numbers, DEA numbers, real addresses.

- Yeah. They're not always fake actually. Costs what, 20 bucks to set up a tax ID number in Delaware, something like that. So we have folks that we call serial incorporators. We get to know their name very, very well because they come in, we catch 'em, they kick off a brand new tax ID number, they come in again, we catch 'em, et cetera, et cetera. To your specific question for that subset where it is fake, we talked about demographic risk models. Those models are looking at things such as, at the time that Dr. Smith this claim, were they actually alive? 'Cause sometimes they were not, right? We also have information we can track with the member. Why is it that you have received seven braces in the past three months? What is going on here? You know. and that could be they didn't necessarily steal your data, steal your information, but there is a call list going around with your information on it, and they'll parse that out and you'll get calls from all kinds of people until it goes stale. And then, I mean it's just like every other telemarketing company, right? So does that...

- I was asking more like how does it affect your relationship with existing practitioners? Because if you blacklist somebody, right?

- Oh yeah, yeah.

- And now the other person is not who they were.

- Yeah, so we don't take that.. I gotcha. We don't take that step on something where it's an existing practitioner. For example, we've had times when we call a existing provider and say, Hey, you know, Mario sending in this claim in a weird way. And their answer is, what are you talking about? But if it does come in under their current tax ID number, that's been doing business with us for a while, it does not come through EPV because we're not setting up a new relationship. So it has to be a new tax ID number, even if it's for an existing provider, and that new tax ID number not used by the existing providers, something we can shut down. And in discussions with the existing provider, they are very much appreciative, right? We've kept them secure. If it's something existing and they've got a bad actor, like a janitor working for 'em that, you know, let's say hypothetically took pictures of the checkbook and hypothetically tried to open their own bank account and et cetera, et cetera, that goes through different meetings.

- [Speaker] Alright, thank you.

- Thank you.

- Hi. So did you have any false positives from your system regarding any provider, where the system said he's fake and then you turn out and go back and find that he's actually a genuine, and if you did, how did you handle, retake that data into your system?

- So if we find a bad actor or provider who fails to authenticate with us, how do we flip that around? So within any fraud program, you know, we're not above the law. We have to give providers the right to appeal that information. In some cases they do. And if that occurs, we can always remove those flags or take care of that information. In the event that we take those administrative actions where we're rejecting them, anytime the provider will submit a claim, they're gonna receive a notice from us that says, "Hey, we don't know who you are. If you wanna tell us who you are, we'll be happy to put that information in and you can start billing your merry way through the company." But yeah, we'll always allow an ability to go back and revisit the work we do. In many of the cases we come across, especially within these DME schemes, you know, they're tied to services that are not being rendered. That doesn't mean that the provider is not set up, they're not licensed, they're not legitimate providers, they're just not providing the service. So the burden of proof is on them now to really come to us and show us, hey, I'm doing what I'm saying I'm doing, or yes I did wrong in the past, you know, let me get better. So we'll always take those actions and wanna work with the providers. Gotta believe everybody's good at heart.

- Yeah.

- As much as I've done this for my entire career and I'm glass half empty all the time and I can't even... I go to the same dentist I've gone to since I was 12 years old 'cause I'm afraid everybody's trying to defraud me. So I have to like put myself in the mindset that sometimes, you know, the glass is half full instead of half empty.

- Yeah. So these are some of the more, I guess, blatant, egregious examples you guys have given. But what about research that you're doing on things that are more subtle. Like upcoding that maybe where the documentation doesn't support that level and like, you know, perhaps millions. Is that something that you guys track as well? And if so, how prevalent is that versus these work groups?

- It's very prevalent. I would say though that's much more of the traditional fraud investigation where we're focused on the who more than the what. We're really concerned with who you are, are you who you say you are? And then what you're doing, we'll kind of put that secondary. We address it, but within the process that this EPV process that really becomes a secondary or tertiary thing that we look at. An outcome might be, we might be looking at a provider, have concerns about their billing and we'll say, yeah, we've got concerns. We're gonna throw that now down, we'll shift it back to the right a little bit. We're gonna throw that to the prepay investigation, or we're gonna throw that to a tip to our SIU to do a post-pay investigation somewhere down the line.

- Gotcha. So your scope is more narrow in that sense.

- Yep. Very narrow, very narrow on the who, but we'll pull every lever. That's the fun part too. We gotta pull every lever in the company within our fraud program. So we'll take advantage of any way we can to stop it.

- [Speaker 2] Right. Okay, thanks.

- Good question.

- So in the presentation you mentioned like the demographic risk models and behavioral risk models, right? So are those built in on Pega or like those are some market tools you are integrated into your workflow process?

- Those are homegrown tools that we've integrated. So yeah, what we did when we stood up Pega very, very quickly, we were more concerned about, does it allow us to move things where they need to go? Does it allow us to manage the documentation? Does it allow us to make sense of the 57,000 tax ID numbers we see, right? So in conjunction with that, we created demographic and risk models that we integrate in. And now with some new capabilities, we're looking at, you know, closer integration and more full utilization of Pega capabilities.

- So in that approach, the workflow piece is Pega, is that correct? Like remaining all or like some external tools you are integrating into this workflow model. Okay. Yeah, thank you.

- Yeah, I think we're scratching the surface on how we can utilize Pega more. Kristen touched on an associated cases button we put in there. It seems so simple, but we have over 300,000 provider records in our Pega workflow solution with all sorts of demographic information. I've got 50 staff members who look at those 57,000 tins. They're not gonna remember a phone number and address they saw in case number one. So from an analytic perspective, we, you know, analytics can be a loose term on it, but that to me, from an investigative perspective was huge in what it brought to us and that was something we couldn't do anywhere else.

- Absolutely, that was a huge step in the process and the program overall. And now with this, our research screen redesign that we're doing where we're breaking down all the data elements that these investigators are gathering on a daily basis, we'll be able to associate on even more granular level of data. So it's gonna be huge.

- [Speaker 3] Yeah. Thank you.

- Thank you. You got three and a half more minutes. Anyone else?

- I'll give you one more question.

- Oh, one more question.

- Yeah.

- Given the scope of that UnitedHealthcare covers, you did a great job of talking about DME examples. What are the plans and the roadmap to cover commercial, Medicare, Medicaid, medical, behavioral, vision, dental.

- It's tomorrow.

- You guys process a lot of claims and do you have the ability now to go in even faster cycles to cover more and more of the scope of the claims you get through your door?

- We were gonna get that done yesterday, but we had to be here to speak.

- Yeah.

- So we have a fairly sophisticated rules engine on the front that tells us what will come in and what is is kept out. Going back to kind of Brent's control chart that he shared at the beginning. That's when you saw that big, oops, that was us trying to do something new, bring something else in, and it did not go well. That was a stressful time. But what we found though is, and this is where again, I think how we leverage Pega is gonna be so huge because when we first started with Pega, I'm an investigator. I needed a box to collect information and I got a box to collect that information, but nobody told me what else I could do in the way that I managed that information, let alone no one really explained, hey, you know, if these five criteria are met, I don't even need somebody to look at it. It can just go down to the next phase. That's gonna be the key to our expansion. So we are gonna be our north star, our blue chip is that every provider within UnitedHealthcare, be it new, known, unknown, will be risk assessed at some point and be risk assessed regularly. And some of the enhancements we're making, I think we're gonna get there probably sooner than we thought too. But I dunno, Kristen. do you agree with that?

- Absolutely, Absolutely. Yep.

- We're eager to get beyond non-par. We're kinda getting bored with non-par. We want to get into credentialing and we wanna get behaviorals been really interesting. Apparently a lot of fraud in the state of Indiana. I don't know what's going on there. Oh, Brent's from Indiana. I didn't mention that, did I? Yeah.

- Yeah. Nothing to see. Yeah, we're fine.

- Cool. This was fun. Thank you guys.

- Thank you.

Tag