At a recent HealthLeaders Exchange, Revenue Cycle leaders asked what exactly is AI, how it works, and how it can be used in the revenue cycle. We have answers.
AI is touted as the latest, greatest advancement in healthcare. But revenue cycle leaders are more than a little skeptical of AI. They're also frustrated, annoyed, and cynical.
"Meaningless," "scary," and "shiny object" are just a few of the ways revenue cycle executives described AI at the recent HealthLeaders Revenue Cycle Exchange last month.
They said they hear lots of sales pitches and hype, but not much about data. They hear about possibilities and promise, but not about real examples of its practical applications, they said.
And, crucially, many revenue cycle executives said they don't fully understand what AI is, frequently confusing the term and using it interchangeably with other technology solutions.
That's why HealthLeaders asked Matt Hawkins, a revenue cycle AI industry expert and CEO of Waystar, every question revenue cycle leaders have wanted to ask about AI.
HealthLeaders: What is AI? How is it different than other forms of computing?
Matt Hawkins: When we talk about computing, in the traditional sense, we're referring to programs that obey a set of predefined rules and logic. A conventional computer can only do tasks that you explicitly program it to do.
On the other hand, a program that runs on AI is designed to mimic the functions of a human brain. Rather than simply obeying commands, software powered by AI has the ability to learn as it goes, identifying patterns and solving problems like a human would.
HL: What is RPA?
Hawkins: RPA, or robotic process automation, refers to software tools that automate human tasks that are rule-based and repetitive. RPA can record tasks performed by an employee on their computer, then perform those same tasks on its own.
HL: What is the difference between AI and RPA?
Hawkins: A simple way of putting it is that RPA mimics human actions, and AI mimics human cognition. Robotic process automation requires a user to perform a specific, repetitive set of tasks. Once the RPA software has recorded this process, it can mimic the user's actions to take over the process on its own. RPA can perform extremely complex processes, but it can't do any tasks it has not been explicitly instructed to execute.
Artificial intelligence, meanwhile, is designed to be as flexible and adaptive as the human brain, learning over time. AI software can interpret vast amounts of data, provide actionable insights, and assist in making decisions.
HL: Revenue cycle executives don't understand AI and they're already feeling hostile and skeptical of it. I have heard them describe it as a meaningless buzzword. One exec has even told his employees he doesn't want to hear the term. Why should they think differently?
Hawkins: Healthcare administration still lags behind in technology adoption. While industries like banking have utilized artificial intelligence for a long time, the revenue cycle still relies largely on manual processes.
I think that because we don't have a clear picture of what AI looks like in healthcare, it leads to misconceptions on both ends of the spectrum: you have some people who fear AI will replace humans because it does too much, and others who are disappointed in the functionality and think it does too little.
In reality, AI is a powerful tool that assists humans with better decision-making and has enormous potential to cut costs and increase effectiveness across the revenue cycle. When you look at the real value that numerous healthcare organizations have derived from using AI to help improve billing and administrative tasks, it's a no-brainer.
HL: What are some real-world ways AI can be or is used in the revenue cycle?
Hawkins: There are opportunities for providers to use AI to optimize every step of the revenue cycle management process. One huge opportunity for artificial intelligence is in predicting claims denials. Providers face the difficult task of minimizing denials from payers, while still processing claims fast enough to keep the practice running. Without insight into the likelihood of denial, provider teams often waste time working on the wrong claims.
What AI can do is predict denials with a high degree of accuracy and precision and build that into the workflow prior to claim submission. By learning overarching patterns and probabilities of claim denials, AI can guide humans on where to focus their efforts in order to maximize the amount of payment received.
After a claim has been submitted, the next step for the provider is to follow up with the payer to settle the claim. Artificial intelligence can help here, too, by interpreting prior history to determine how long it will take a specific payer to settle a claim. AI tools can show, statistically, when a claim has gone unpaid for an irregularly long time and requires human intervention. Again, this increases efficiency for healthcare administrators, helping them manage their time so they can direct their efforts to more important tasks.
AI is also a valuable tool for ensuring a better patient financial experience. As patient financial responsibility continues to grow, it is crucial for providers to provide a seamless, consumer-friendly billing experience while safeguarding a healthy revenue flow. AI tools can interpret data to model a patient's propensity to pay, and then offer insights on how to send the right follow-up message at the right time for that patient.
AI can also help determine whether a patient is eligible for charity care, saving money for hospitals and patients alike.
HL: How are those applications different than something like automation or other forms of rev cycle technology?
Hawkins: Take the example about predicting claims denials. Robotic process automation tools allow providers to automate the claims denial process, which is helpful for reducing manual effort and minimizing errors.
However, AI takes this a step further by collecting and interpreting data as it goes, and then using that knowledge to continuously tweak and improve the process. AI offers insights and ideas for improvement that RPA, which functions on rote repetition, cannot.
HL: A lot of revenue cycle executives seem to be taking a “wait-and-see" approach to AI. Is this the right strategy for 2020?
Hawkins: Between heightening operating costs and difficulties collecting patient payments, healthcare providers are under more financial pressure than ever before. At the same time, medical billing faces heightened scrutiny nationwide, particularly around surprise bills. These factors are not going to mitigate in 2020. It's more important than ever for healthcare organizations to ensure that their billing practices are as accurate, efficient, and straightforward as possible, and AI is the most powerful tool available to achieve that goal.
The HealthLeaders Revenue Cycle Exchange is one of six healthcare thought-leadership and networking events that HealthLeaders holds annually. Our Revenue Cycle Exchange allows you to share insights and ideas with other revenue cycle VPs and leadership with the same challenges. To inquire about attending the next HealthLeaders Revenue Cycle Exchange program at the Omni La Costa in Carlsbad, CA, April 20-22, email us at firstname.lastname@example.org.
Alexandra Wilson Pecci is an editor for HealthLeaders.
AI-powered software learns over time, identifying patterns and solving problems by mimicking human cognitions.
Robotic process automation software tools automate human tasks that are rule-based and repetitive.
AI revenue cycle applications include predicting claim denials, following up on submitted claims, and customizing patient financial encounters.