How Is Robotic Process Automation Evolving in Healthcare?

Last fall consulting firm Baker Tilly bolstered its robotic process automation (RPA) capabilities with the acquisition of advisory and implementation services firm Alirrium. In a recent interview with Healthcare Innovation, David Hickey, principal with Baker Tilly’s digital solutions team, spoke about how RPA’s use is evolving and accelerating in the healthcare sector. 

Healthcare Innovation: First, can you define robotic process automation and how it has been evolving over the last few years? 

Hickey: Robotic process automation is a low-code technology that allows you to build digital workers. These digital workers can simulate anything a human can do on a computer screen. If you can pull it up on a computer and interact with it, my team can build a digital worker that can simulate those actions. It is just executing those highly repetitive rules-based processes. Anytime there’s data that has to move between applications or systems. that’s when you use this technology. 

Over the last three to four years in particular, they’ve started integrating them with machine learning engines and now agents, so they’re getting smarter. The machine learning engines do things like take any document, identify what that document is, and then, based on that document type, intelligently extract the data from the document, regardless of the format. The easiest one to explain is actually invoicing, because every invoice that comes in from a different customer has a different format. So the machine learning algorithm would first identify it as an invoice and based on the invoice type you would train that machine learning engine how to extract the data.

HCI: In the healthcare space, what is some of the low-hanging fruit that this has been applied to — things like claims processing?

Hickey: One is good faith estimates. You might log in to the payer’s system, and using all the information that the patient’s already provided, you would submit all that and you would get information back, and then you can submit that back to the patient. That would be something that is mainly done manually today.

Healthcare groups have done a fantastic job of tracking, tagging and managing all this data, but now they’re drowning in this digital assembly line, as we call it. In a lot of cases, it’s a portion of a lot of people’s jobs that we’re relieving from them. It is like digital assistants helping the humans, so the humans can actually do the things that are intrinsically human — collaboration, using their creativity, communicating with clients. It’s refocusing them on higher-value-added activities instead of doing this stuff that they’re not great at it and it’s very monotonous.

HCI: Is this happening mostly at large health systems? Or are mid-sized or larger physician practices also getting on board with automating a lot of the administrative or patient outreach tasks that they normally have people do?

Hickey: I would definitely say the ROI is generally driven by volume. These robots, they run 24/7 doing good faith estimates during the day, and then doing claims processing at night. So the no-brainers are the big hospitals, because of the high volume and the impact it’s going to have. Today, you can deploy this technology and it has a big payoff for them. But we’ve had plenty of mid-sized clients, too, because, depending on the size of the practice, you know, there are certainly use cases for going in there and building bots that that would support your rev cycle within the overall process.

HCI: For Baker Tilly, is there a vendor landscape of tools you use or are you creating something from scratch?

Hickey: We’re using tools. There are a lot of RPA applications. UiPath is one that we’re experts in, but we’ve also used Power Automate. There are probably 20 or 30 different major RPA vendors.

HCI: Would you say that use cases in healthcare are relatively more complex than in other verticals, or fairly straightforward?

Hickey: I think they’re complex just because your rev cycle is so complex, with all the codes and the buyers and payers and authorizations, and it’s different between different payers. Certainly, we have a lot of customers in different industries where it is much easier to implement the technology. 

But I think the major selling point of the technology is the speed with which you can develop these automations and the flexibility. It can just mold to whatever your process is. I used to do a lot of large-scale ERP implementations. When you would deploy the technology, you had to change your business to fit the software technology and how it was designed. And if you deviated too far from the core design, you had to do all this custom development, and it was extremely expensive to do that, and then basically you can never get rid of your consultants. The true power of RPA is that I literally sit down with you and your team and say, ‘show me how you do your job.’ And we build out the as-is process. We understand exactly what you do today, and then we ask what would this look like in a bot-enabled world? We’re slightly changing some of the things that they’re doing to make it more efficient, but it’s really just conforming to whatever your process is. 

HCI: Are the payers organizations getting on board just as much as the as the health system side?

Hickey: Yes, they are, and I attribute it to the versatility of the tool. Everybody has back-office business processes that can support this technology. 

It is interesting, we’ve had customers come to us and say, from a security standpoint, we’ve hardened our outer shell from the world. But we want to build bots that interact with our critical HIPAA data, or our Personally Identifiable Information data, because we want to limit access to this. Robots do exactly what you want them to do, and they don’t deviate. And if anybody tries to modify them, they break. Then you have to bring them back to the core development tools to make changes to them. So it’s been kind of interesting. I’ve had a bunch of clients say that was their primary driver. It wasn’t ROI, it was security. It was a little surprising when I first started hearing it, but I’ve heard it a lot over the last eight years. 

HCI: Do you have to have humans monitoring the robots to make sure that that quality assurance is is always where you want it to be? 

Hickey: So you’re going to have what we call humans in the loop. You have “rumble strips,” which define the parameters for where your bot can operate, especially when you start to get into the agentic side of things, where it’s starting to make decisions for you. Once you hit a rumble strip, that’s an area where the bot’s going to say, ‘Hey, I’m not allowed to operate outside of these bounds. I need help.’ Then it would shoot off the exceptions for the humans to manage. It can integrate with you and work with you along the way. The other neat thing about that is every transaction is tracked, so you have a new level of visibility, because we’re tracking everything in real time at the transaction level.

HCI: When you engage with a health system, who is it that you’re most likely dealing with? Is it the CIO, the CFO, or revenue cycle leaders, or a committee of all those people involved in the decision-making?

Hickey: It depends on the organization, but yes, all those people that you’ve mentioned. Obviously the CIO at some point has to be involved, because we’re deploying a software technology, and they’re going to need to spin up some virtual machines to load robots, because these robots have got to be on your network, and we have a tool that credentials them so it has all the encrypted credentials, because that robot is essentially acting as a human. It’s logging into applications, just like a human.

HCI: Do the clients give them names? 

Hickey: Oh, yeah, they give them names. I have one customer that has 18 different robots, and they’ve named all their robots different fun names. They come up with themes, like Star Trek or the Avengers or something like that. One time we deployed a robot, and one woman who wasn’t part of the deployment team, but was the recipient of the robot’s work, wrote her boss and said, ‘Hey, I think I’m going to divorce my husband and marry Bob.’ Bob  was the name of the robot. 

HCI: If health systems have not dipped their toes into RPA, are they probably getting nervous that they are not achieving that level of efficiency that other people are?

Hickey: Everybody’s under pressure to drive efficiencies, so the use of digital workers is going to make more and more sense moving forward. The technology has been around for 20 years, but the acceleration we’re seeing right now — I haven’t seen anything like it before. We usually don’t do implementations unless the ROI is less than a year, and it’s usually something like six months. I tell people, it’s sort of the gateway tool to AI because there is an agentic aspect of it. So you can train a persona for a robot and say, you’re an AP clerk, you’re processing invoices. Here’s your job and here’s the context, here are the standard operating procedures and the databases that you have access to. You create these personas, but they’re very targeted. You’ve got to make sure that their job is very well defined. 

People are getting much more comfortable with it, although I would definitely say that there’s a general sense of unease about the speed at which technology is changing. I think things are changing so fast that the human adaptability curve is now below the technology curve. So there’s a general unease, because people don’t know where it’s all going.