Integrating artificial intelligence and automating workflows can significantly improve hospital operations, particularly in revenue cycle management, and many providers are developing AI tools to help with third-party payer denials and the rising cost of collections, among other areas.
About 74% of hospitals and health systems are actively automating some portion of their revenue cycle operations, with an additional 46% of these organizations reporting using some form of AI, according to a recent report published by Akasa and the Healthcare Financial Management Association.
While workers' fears about AI taking their jobs is softening, the onus is on health system leaders to ensure their team understands what AI means for their role, what automation is and what it is not, and how it unburdens them from some of their repetitive, administrative tasks on a day-to-day basis.
"People are starting to turn the corner in thinking about AI as a threat to what we do on a day-to-day basis. A lot of it is on us as leaders to show our team we have the ability to do five other things we may not have had the time to get to previously because the technology has removed a lot of this other administrative work," Aaron Lewis, CFO of Brentwood, Tenn.-based Lifepoint Health, told Becker's. "It's the same with nurses at the bedside: if we can take away five things they have to do through automation, there's no doubt there are higher impact, higher quality, patient-facing activities they can spend their time on instead."
Health systems are beginning to see a shift in their employees' mindsets when it comes to AI, but to ensure buy-in, it's critical for leaders to explain how they are using the technology to augment their workforces, not replace them.
"It's about taking those things away that are probably the lowest value, highest friction processes, and positioning people to be more impactful in other areas," Mr. Lewis said. "That's the pivot people are starting to make, but it also requires human adoption, which is hard. It's on us as leaders to articulate how we're repositioning people and helping them understand how to leverage the technology working with them and for them to then reshape what they do on a daily basis. Change is hard, but it is also critical for us in this space."
Hospital RCM teams that were early adopters of AI, or those that are only beginning to implement AI tools, have encountered pushback and concern among some workers. However, there is no shortage of work in the revenue cycle, and the work is becoming increasingly complex, according to Amanda de los Reyes, vice president of revenue cycle at Valleywise Health in Phoenix.
"One of the things you can do to quell people's fears is make sure they understand what the solution is doing and how it works, then follow up by explaining how they can work on [another project]," Ms. De los Reyes said during a panel discussion at Becker's Revenue Cycle Management Virtual Event. "It's a great way to help people get a little more comfortable. Those who are uncomfortable or don't understand the technology will push back and ask questions about what this means for their job, following these two steps ensures a lot more success with the implementation process."
Dennis Shirley, vice president of revenue cycle at West Des Moines, Iowa-based UnityPoint Health, said that engaging your teams early in the conversation and showcasing early AI wins also helps ensure buy-in.
"First and foremost, engage your frontline team transparently and early in the conversation. We made sure our teams understood what automation was and what it wasn't," Mr. Shirley said. "We talked about the fear factor, which is the first thing that comes to mind for a lot of people asking, 'What does this mean for my job?' So, we intentionally worked to find some early wins that would assist the teams in their biggest headaches."
An example of this was revenue cycle workers having to log in to a certain payer's website multiple different ways for each of its application programming interfaces (APIs), which could take about 12 minutes.
UnityPoint automated this login process to save workers time and highlight an early win.
"Quick, early wins like that show that it doesn't all have to be us automating the assembly line. We can automate these small processes and systems and the human-bot/human-AI interaction continues to be necessary," Mr. Shirley said.
UnityPoint, which includes 20 regional hospitals and 19 other community network hospitals, is automating transactional, day-to-day activities and utilizing better insights to help the health system pinpoint where it should be focusing its resources.
"AI as a big theoretical concept is sometimes difficult to digest, so we're trying to break that down to a micro level," Mr. Shirley said. "What can we automate? What are some of the redundant, repetitive tasks that make sense from a [robotic process automation] perspective plus where are there opportunities for us to utilize insights on big data modeling to help us increase efficiency in the way that we do our work?"
One key area in which UnityPoint is using AI is by looking at its credit balance management and identifying true credits (meaning refunds) versus credits that are over contractualizations or credits that do not need manual intervention.
"We're able to use historical data to predict that these are the types of credits that do require intervention, and queue them up for our team members to tackle," Mr. Shirley said. "These are the types of credits that need contractual adjustments to reconcile and move on, with the aim of making that team and their finite number of resources and touches available hit the right things to ensure we're making those touches as efficient as possible."
Another big revenue cycle opportunity is to incorporate an AI algorithm into health systems' denial prevention strategy,
"The best denial strategy is prevention," Ms. de los Reyes said. "We're proactively looking at where we might see an uptick in a certain type of denial or even with a certain payer. This helps us know in advance — instead of a couple of months down the road — that we have a bunch of denials from a particular payer. You can start to work with the payer a lot sooner if you notice some of these things. The same goes for managing [accounts receivable]: we used some of these strategies during the Change Healthcare issue to help us identify some of the payers that were struggling because we didn't use some of those tools, but some of our payers did."
An ounce of prevention is worth a pound on the back end of the revenue cycle, according to Mr. Shirley.
"If we can reallocate resources that otherwise would have been spent doing transactional activities to denial prevention, we can ultimately reduce leakage and improve our yield," he said.
Revenue cycle leaders cannot overstate how critical it is to ensure that each AI function is working properly, producing the outcome that a health system is looking for, and that leaders are continually auditing the automated processes.
"Once you get the automated processes going, you need to ensure you're doing a PDCA (plan–do–check–act) to make sure it hasn't gone off the rails because it can spiral very quickly when that type of issue happens," Ms. De los Reyes said. "Often our EHRs are getting different types of upgrades so we need to make sure that the embedded AI we have flowing is still working properly. Test, retest and retest again, especially as you loop in with other products and solutions instead of just your EHR. Make sure it's functioning from end to end because that is one of the issues that can cause some team members to become hesitant about wanting to use AI functions or tools, when they're actually very helpful."
Revenue cycle teams need to continuously audit and manage what they are automating, and be selective about what they are automating in the first place.
"Make sure that you're automating a good process. Automating a bad process just replicates the bad faster and more. You also can’t just set it and forget it; you have to continue to manage your automation, similarly to how you would manage your team members, whether that’s related to payer changes, system changes or even just normal evolution of the work. And that is a leadership change," Mr. Shirley said. "When we talk about leadership development and the impact of the rise of automation and AI, our leaders have to change how they interact. It's a different skillset necessary to manage the growth in AI and we need to make sure we're investing appropriately in our leadership acumen to be able to deliver that."