Artificial Intelligence

AI, automation and RPA in the revenue cycle

February 1, 2021 8:21 pm

As 2020 threw a curveball at healthcare, leaders identified revenue cycle as the area most ripe for innovation, an October 2020 survey found — and 57% of survey respondents are confident that innovation in revenue cycle can happen in the next year. How will artificial intelligence (AI), automation and robotic processing automation (RPA) fit within the revenue cycle of the future, and what steps are hospitals and health systems taking now to design the revenue cycle of the future? In this roundtable, conducted during the HFMA Digital Annual Conference, 10 revenue cycle leaders share insight.

Describe your organization’s revenue cycle technology maturity journey, particularly as it relates to RPA, predictive analytics, to the adoption of artificial intelligence.

Amy Assenmacher, senior vice president, revenue cycle, Spectrum Health: “I would say we’re mature, especially in terms of RPA. We became more serious about RPA 10 years ago when we partnered with a vendor that was proficient in this space, and we identified many use cases to start with, especially over the past couple of years. We’ve also developed an RPA placemat of sorts, which is an inventory of our RPA use case processes. I do think we have some opportunity in terms of intelligent automation, progressing to more complex use cases. Yet, we’re very proud of the work we’ve done to date.”

Randy Gabel, senior director, revenue cycle, OhioHealth: “For the past year and a half, we’ve been doing some work around automating claim statusing. We’re on a three-year automation plan that includes many different things, from RPA to AI to predictive analysis. We’d like to get to a point where we can predict denials and resolution.”

Krishna Tummalapalli, portfolio architect, OhioHealth: “I’m an enterprise architect, and I advise the chief technology officer. As Randy was saying, our automation journey has started, and we are currently looking at claim statuses and prior authorization as our key use cases to explore in 2021. When it comes to predictive analytics and AI, as an enterprise, we leverage them for some processes.”

Robert Dewar, vice president and chief revenue officer, Geisinger: “At Geisinger, we’re in an interesting situation in that our fundamental revenue cycle technology is from the 1980s, and we’re just moving off of that technology onto an integrated Epic platform, front to back. That’s an interesting journey in itself. We’ve still managed to adopt RPA across a number of functions, and we fully expect that having Epic will provide some additional views and ideas around RPA. Right now, we have about 19 bots running, with several dozen more in the works. In terms of predictive analytics, we rely on outside vendors and their analytic tools, particularly around patient pay and the best way to encourage patients to pay their out-of-pocket portion, which is extremely important in this environment. We’re just dipping our toes into AI as well.

“It’s a little bit of the ‘Wild West’ out there: Everybody’s talking about AI, and there are not a lot of places yet that are really doing it well. We have a company that is performing autonomous coding in the emergency department for us, and that’s our initial foray into AI, but we’re also looking at other applications, such as on the back end around insurance follow-up.”

Lucas Foust, senior director, revenue cycle, UPMC: “UPMC was an early adopter of RPA through an internally developed solution that has been in place for well over a decade. We’re in the early stages of evaluating a third-party RPA solution, the benefit to which would be that our current solution isn’t integrated into our legacy revenue cycle solution. And as we go through an Epic implementation, we see a lot of opportunity for this third-party system to integrate with Epic. In terms of predictive analytics and AI, the vast majority of our historical and current analytics have been very descriptive in nature. My team is investing in training right now on cloud solutions that would hopefully accompany a transition into more unstructured data. This would help us bridge the gap between prescriptive and predictive analytics and convert some of our RPA solutions into more of an AI-driven solution using machine learning as a brain power to get us there. We’ve evaluated a lot of AI solutions in the market, but we really haven’t found something that meets our needs.”

Terrie Handy, vice president, revenue cycle operations, Legacy Health: “We’re in the process of teeing up an automation, AI and RPA strategy and charter. We’re first reassessing Epic automation: What is our current state? What is our inventory? Where can we have additional enhancements within that system, from an automation perspective? We’re doing this via an Epic refuel, but we know Epic can’t do everything. We’ve also rolled out key initiatives, from automation to predictive analytics and self-pay segmentation. I’m also partnering with key leaders both within and outside of Legacy in an effort to help develop an overall strategy that we can implement relatively soon, even if it involves pilot or trials as we work toward a longer-term solution.”

Mary Beth Remorenko, vice president, revenue cycle, Partners HealthCare: “We incorporated RPA into revenue cycle three years ago, and we worked with a vendor to develop our own system-level department — the Department of Intelligent Automation — to prioritize RPA opportunities across departments, including revenue cycle. Within revenue cycle, there is an analytics group that evaluates opportunities for RPA and monitors RPA initiatives to make sure they are yielding either the cost savings or revenue improvement that we anticipated. Unfortunately, it hasn’t been as much of a cash cow as we would have liked, but we have done some things that have resulted in incremental value, such as through denials follow-up, referral submission and notice of admission. I was able to cut a million dollars from my budget for the upcoming fiscal year based on our work in this area.”

Mark Norby, revenue cycle chair, Mayo Clinic: “Our revenue cycle leaders have a grand vision for automation. We want to automate as much as we can, and we want to use the right tool at the right time for the right price. We are utilizing Epic’s automation to its fullest. Our strategy is to use that base system that we have already paid for and then bolt onto it with RPA and other forms of automation, like machine learning and AI.

“From a revenue cycle standpoint, our use of predictive analytics is new. We’ve been tasked with doing what seems like is nearly impossible, but we’re going to do it anyway, and that is to predict gross and net revenue into the near future. That’s really where we’re spending time with predictive analytics.”

Stephanie Wells, system vice president, revenue cycle/HIM, Ochsner Health: “We’ve been doing claim statusing for about five years now, but we really dug into RPA about two years ago. We have a partnership with an AI software vendor, and we’re an alpha site for them. Right now, the only place where we truly deploy AI in the revenue cycle is around our capitated population with our [hierarchal condition category coding] work to ensure we’re capturing risk appropriately. We also use predictive analytics for risk segmentation, but we have a long way to go in using predictive analytics in revenue cycle.”

What revenue cycle processes in your organization would most benefit from the use of innovative technology such as analytics or automation? Consider the patient experience and their flow through from pre-visit to payment (zero balance).

Wells: “A lot of our focus with RPA has been on the authorization side — not only statusing authorizations, but also checking to see if a service requires authorization through RPA and actually submitting information for the authorization through RPA. That’s huge for us to save a lot of work there. We’re also moving our claim statusing to RPA not only from the hospital side, but also the physician side. We hope that the learnings from these efforts will inform our work around predictive analytics.”

Handy: “From a RPA perspective, I think anything we can do to streamline the patient visit well ahead of time, such as prior authorization, enhances the patient experience. I wish we could streamline the entire process with our payers as I believe we’ve shown that some of the authorizations are not necessary, but that’s another conversation. As we move toward compliance with the new price transparency rule, many of us are working to implement a patient estimator tool, and there will be some automation involved in that. Additionally, automating the notice of admission has been one area of focus for us. Approximately five years ago, we turned on a direct notification with United Healthcare for notice of admission. Several of our key payers, including a large national payer in our market, have stated they are not ready for this automation or the 278. That said, we continue to push for this automation.”

Tummalapalli: “Document sharing is another area that would benefit from automation: There are documents that we’d like to share proactively, and there are others that we would like to share on demand. Creating a platform that can assist in exchanging this information securely and provide exactly what is needed will be a big-ticket item.”

Remorenko: “We’re working on automating price estimation and leveraging a combination of Epic capabilities to do that. We’re also identifying the types of analytics that can help inform our pricing strategies so that we can be more transparent with patients about their out-of-pocket liabilities.”

Dewar: “One new area that is ripe for automation during COVID-19 is around patient arrival: using automation to support a touchless arrival so that the patient is arriving in a safe way and then capturing all of their information for bill payment in almost an airline kind of fashion. That’s an area that we’re working to perfect right now.”

Describe your organization’s readiness to adopt new tools like automation, RPA or AI. What are challenges for adoption of new tools? Where are your biggest pain points in your revenue cycle processes?

Assenmacher: “There’s a lot of really good energy here around RPA and in creating that center of excellence across the organization so that efforts are coordinated, but we don’t yet have a cohesive RPA strategy for the system. Within revenue cycle, the challenge with RPA specifically is that there is resistance from some payers to allow that kind of screen scraping by providers.

“When you’re developing scripts for RPA, you have to be really mindful that if you go through a system upgrade, you may have to rewrite the scripts. Some bots are really sensitive to those changes, and that can have a cascading effect, so it’s also important to be aware of those dependencies. Additionally, when you implement RPA or AI, make sure you understand how the whole system is connected, and establish good oversight and monitoring to ensure that the output is accurate and that these technologies are executing in the way you’d like. It’s also important to establish metrics for automation, such as the impact on patient experience scores, the percentage decrease in denials or improvements in prior authorization.”

Dewar: “One of the advantages that we have at Geisinger is that the CFO is a very strong proponent of everything from RPA through AI. He’s continually looking for better, cheaper ways to do things. We can’t just keep adding labor to revenue cycle as the complexity of reimbursement grows, so it’s very useful to have that kind of pressure from the top to gain alignment across the organization.”

Remorenko: “One of the RPA challenges we’ve faced has been workflow readiness. Some of our revenue cycle workflows were not documented, so we’ve spent a lot of time during the RPA evaluation process in documenting workflows. It’s very similar to a process improvement project, determining all the steps involved, and we’ve had some lessons learned along the way, such as when there are 10 exceptions that we hadn’t known about. We’ve learned a lot about the importance of process mapping and evaluating all of the steps involved in our revenue cycle workflows.”

Tummalapalli: “It’s also important to have a strong software engineering team as well as a strong data science team as you explore analytics, AI and process automation. Obviously, partnerships with vendors that have maturity in this space are very important, but as Mark was mentioning earlier, if you want the cheapest option, you should also be maturing as an organization so that you can maintain and strengthen your in-house capabilities. Randy and I, along with our leadership, were really successful in making automation an enterprise priority, establishing an enterprise (not just for revenue cycle) center of excellence. One key thing we noticed is that when people have been working in the software engineering and data space for a longtime, they also naturally become knowledgeable about the operational areas they support. When they are involved on a project, they tend to have great ideas on what else can be achieved using the same technologies and solutions. From an organizational readiness standpoint, that’s key.”

Norby: “At Mayo, we’re focusing on process engineering and process reengineering, and that’s how we believe we’re going to get to selecting the right tool at the right time for the right price. We feel like we’re really ready for this transformation. Our biggest pain point is that multiple vendors are in this space, including new entrants, and everyone says what they can do, but very few can prove what they do. Our biggest struggle is recognizing what they sell versus what they actually do.”

How would your organization define success in implementation of new tools and processes in the revenue cycle?

Remorenko: “Our team created a metric called ‘capacity created,’ measuring the amount of time and effort that was saved when RPA went live. Currently, we are tracking this metric on a weekly basis for each bot.”

Wells: “We take a very similar approach, looking at how much work the bot is taking on that previously would have required a number of FTEs and the cost that is saved through use of the bots. We have a dashboard that shows us what is happening by bot, and we review the dashboard on a weekly or biweekly basis. We also review our productivity standards for our FTEs and consider what our needs are now that we have RPA. Ultimately, we’ve found that by taking the mundane functions off our employees’ plates, we’re enabling them to work at the top of their license. This improves employee engagement by keeping them interested in their work.” 

Norby: “We’re also focused on maintenance of the bots. One of the challenges with RPA is that it requires controls in place to ensure any changes in process, systems or external websites (i.e., payer websites) are constantly up to date; otherwise, it may essentially bring the bots down, resulting in interrupted operations and requiring hours and hours of maintenance to get them back up. We’re finding that some of the newer bots can be maintained and developed with a lesser skillset. That saves time for the end users — our revenue cycle operational peers — while also reducing the time required to keep the bots up and functioning.”

Gabel: “We track the productivity daily. I receive email updates regarding how the bots are performing as well as alerts if the bot has failed or is experiencing any problems. We can react to those problems fairly quickly. I’m excited that I can review a productivity dashboard daily and see that there’s nothing to be alarmed about; everything seems to be going well.”

How has COVID-19 impacted your capacity to do process improvement? Are you currently working on improvement efforts? What are your revenue cycle priorities during this time?

Wells: “We put a lot of focus on low contact on arrival, trying to make sure patients felt safe. The other thing that took a lot of work during COVID-19 is that payers would change the rules often. We would be in the process of getting everything built to be billed, and then, there was a change. It took a lot of effort to make sure we were monitoring every single payer change all the time; making sure the build was set correctly; making sure we were being diligent in our processing. We always are diligent; it’s just that there were so many rule changes that happened so quickly. We are also tracking all the updates we’ve made to our processes and systems so that we are prepared to resume previous billing rules when the pandemic health emergency rules end.”

Dewar: “I completely agree. It was amazing how many times we’ve touched some of our COVID-related bills. What surprised me about our response was how easily we got people offsite into a remote work environment and how quickly we managed to pick back up on our process improvement efforts. It shows that people were willing to adapt.”

Describe your organization’s revenue cycle governance structure. How are clinical processes integrated into revenue cycle operations? How do you ensure alignment between clinical and revenue cycle operations? How does process improvement take place? What have been successes and challenges in this regard?

Handy: “I’ve been doing a lot of research around this and our CFO is definitely very engaged and supportive. We recently hired a new director who oversees analytics, and at this point, that position is dedicated to population health and our finance and payer strategy. We need a similar position in revenue cycle, and we’re looking at our current budget and examining whether we can reposition a FTE to move into that role and then further develop a team that is focused on analytics as well as automation.”

Assenmacher: “We’ve gone through a couple different iterations post-Epic, trying to ensure we have that robust connection between clinical and finance. Recently, we’ve launched, across Spectrum Health, the SAFe, or scaled agile framework, which is an  innovative model designed to align, organize and synchronize projects and priorities with a limited pool of resources across an organization. This model is essential to the pace at which projects can be executed on. Additionally, the robust huddle structure in place here is very effective in terms of ensuring we’re touching base at least weekly to review key priorities and strategic initiatives. Additionally, our strong project management program is essential in aligning resources and projects across the clinical, financial and operational areas.”

Remorenko: “We definitely have a lot of work to do in this area. A lot of times, our clinical functions are moving forward with one thing, and then our revenue cycle team finds out about it after the fact, and we come in late and have to figure out, ‘How do we bill for expanded COVID-19 testing? How do we do all of these things?’ We’re making improvements. We try really hard to educate the rest of the organization on why revenue cycle needs to be included in some of these clinical decisions or initiatives. But we definitely have a lot more work to do.”

Wells: “We have governance over automation, and we have finance councils by service line where revenue cycle and clinical teams talk through the initiatives that are in play. We also have an area that focuses on new services. However, despite all of this, sometimes we find out something went live after the fact, and we have to back up and get everyone to the table. So even though there’s structure, there’s still the possibility of something happening outside of this structure that creates surprises for revenue cycle as well.” 

To learn more about Simpler’s Revenue Cycle Management Consulting Practice, visit

About Simpler Consulting, an IBM Company

© IBM Corporation 2020. All Rights Reserved. The information contained in this publication is provided for informational purposes only. While efforts were made to verify the completeness and accuracy of the information contained in this publication, it is provided AS IS without warranty of any kind, express or implied. In addition, this information is based on IBM’s current product plans and strategy, which are subject to change by IBM without notice. IBM shall not be re-sponsible for any damages arising out of the use of, or otherwise related to, this publication or any other materials. Nothing contained in this publication is intended to, nor shall have the effect of, creating any warranties or representations from IBM or its suppliers or licensors, or altering the terms and conditions of the applicable license agreement governing the use of IBM software. References in this presentation to IBM products, programs, or services do not imply that they will be available in all countries in which IBM operates. Product release dates and/or capabilities referenced in this presentation may change at any time at IBM’s sole discretion based on market opportunities or other factors, and are not intended to be a commitment to future product or feature availability in any way. Nothing contained in these materials is intended to, nor shall have the effect of, stating or implying that any activities undertaken by you will result in any specific sales, revenue growth or other results. Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon many factors, including considerations such as the amount of multiprogramming in the user’s job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve results similar to those stated here. All customer examples described are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual environmental costs and performance characteristics may vary by customer. IBM, the IBM logo,, Watson, and Watson Health are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at “Copyright and trademark information” at

This published piece is provided solely for informational purposes. HFMA does not endorse the published material or warrant or guarantee its accuracy. The statements and opinions by participants are those of the participants and not those of HFMA. References to commercial manufacturers, vendors, products, or services that may appear do not constitute endorsements by HFMA.


googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text1' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text2' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text3' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text4' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text5' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text6' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text7' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-leaderboard' ); } );