How Artificial Intelligence Can Streamline Revenue Cycle Management

Artificial intelligence (AI) may have once seemed too difficult to widely implement in a department as complex as the revenue cycle. However, a new KLAS report reveals that leaders at top healthcare organizations still think the technology is worth the effort.  

The survey findings indicate that most respondents plan to prioritize automation in the next 12-24 months, with a focus on autonomous coding and claims management. The drive to innovate the revenue cycle through AI-enabled technology remains high due to its potential to eliminate repetitive, time-consuming tasks, boost reimbursement, and reduce denials.  

Despite various obstacles to automation, like cost, resistance to change, and a lack of technological understanding and trust, leaders express a continued commitment to overcoming the challenges to streamline revenue cycle management (RCM).  

Still, introducing AI into the revenue cycle will bring changes for health information (HI) professionals. They may need to identify inefficient coding processes and partner with information technology and AI vendors to update the algorithms for accurate code capture. Their workflows may shift, and they may need to upskill to navigate a new system and job duties.  

Benefits of Autonomous Coding

Autonomous coding programs use natural language processing and machine and deep learning to scan electronic health records and automatically assign ICD-10-CM and Current Procedural Technology (CPT) codes to encounters. The claim then transmits directly to the payer for reimbursement, typically all without human intervention on the backend.  

Once up and running, experts say autonomous coding can quickly reduce billing backlogs and improve coding compliance and efficiency. Increasing the productivity of the existing HI workforce can be a big draw for revenue cycle leaders, who are under pressure to keep operational costs down and cash flowing, particularly as their departments struggle to find qualified candidates.  

Maggie Kenworthy, MHA, RHIA, CDIP, CCS-P, CRCR, director of charge capture and coding operations with McGovern Medical School at UTHealth Houston in Texas, says it makes good business sense for RCM departments to use autonomous coding for simple or repeat claims that have structured documentation, such as radiology services.  

Although HI professionals may initially need to verify more of the automated claims, she says only spot checks are required once the program learns the parameters and can abstract with 95 percent accuracy or above. The current technology is not ideal for every service line, such as surgery or where documentation is less structured, so plenty of work remains for HI professionals, says Kenworthy. However, it's a reminder that the industry is heading in a new direction, and they should prepare.  

"HI jobs will change. There will be less manual abstraction and more quality review," she says. "Coding might actually be a little more difficult, and HI professionals will need to have a slightly different skill set, with a questioning attitude and critical thinking to really apply their guideline knowledge to auto-suggested codes."   

Kristie Janvrin, CPC, associate director of ancillary services for the central professional billing organization at Mass General Brigham, an integrated health system headquartered in Boston, has seen firsthand how an AI application can boost employee satisfaction. She says AI's ability to free HI professionals from low-complexity coding is a win for them and others, including patients. The health system uses code prediction and automation for radiology, cardiology, pathology, and endoscopy, with plans to expand to other departments like emergency care.   

"Skilled coders want to think outside the box and not just do remedial work or click a button to validate, so using AI leaves the higher-complexity cases for them to code," says Janvrin. For radiology claims, they've reached almost 75 percent automation within the first year of implementation. They will keep adjusting the system until quality thresholds rise to 95 percent or higher accuracy.  

Although some HI professionals may worry about job security, Janvrin was excited when her directors decided to pursue AI. Their goal was to reduce billing lag, increase productivity, and better utilize HI professionals' unique skill sets. She says the benefits have only compounded.  

With the time savings, the team can now dig into other issues affecting the revenue cycle, such as incomplete documentation. "Because of AI, we have time to collaborate with each other and actually fix problems, plus do peer-to-peer training and advance coders to another level, which provides them with additional compensation," she says. Many on Mass General Brigham's central professional billing team have gone on to assume quality assurance and compliance roles, providing physician education and determining edits to add to the claims scrubber to prevent denials.  

Drawbacks of AI in the Revenue Cycle

Autonomous coding technology — no matter how advanced — is only as good as the inputs, says David Lane, BS, FHFMA, independent revenue cycle management consultant at Revenue Cycle Performance Group in Chattanooga, TN.  

"When you start getting into more complicated encounters, you have to revise the canned scenarios and have a better system for checks and balances," he says. For example, if an inpatient is treated for pneumonia and dehydration, but AI isn't preprogrammed to pick up the latter code, the organization can miss out on a significant amount of additional diagnosis-related group (DRG) reimbursement, says Lane. A strong auditing program can prevent those unnecessary losses.  

Like other RCM processes, implementing AI requires a long-term commitment and continuous improvement to identify gaps. Kenworthy says the build-up at her institution has been progressive over the past several years. "It has taken more than eight years to work with the vendor, optimize the software, and allow the system's logic to learn and adapt," she says.  

It may also take a few tries to find the right vendor. Janvrin says they previously used two other AI applications but could not exceed 30 percent coding accuracy for automation until they moved to their current system.  

Other Revenue Cycle AI Opportunities  

While autonomous coding has garnered much attention, AI technology can streamline other aspects of the revenue cycle. Lane says that healthcare institutions already use automated systems to verify registration information, collect patient balances, and post payer payments, with minimal staff involvement. He says predictive dialing systems can potentially expedite claims follow-up, especially status inquiries. 

AI technology holds promise for denials management, too. Kenworthy says that generative AI like ChatGPT will likely gain traction in automating appeal letters and requesting more specificity from providers when they're documenting an encounter. Her organization is also exploring a bot to streamline medical record release to payers.  

Experts agree that AI will bring substantial changes to the revenue cycle. In the meantime, Janvrin encourages HI professionals to keep an open mind and embrace it. "Taking away the mundane workload has made our lives a lot happier and the job more fulfilling. Change can be uncomfortable, but empowering, and move you into a more sophisticated RCM role." 

Steph Weber is a Midwest-based freelance journalist specializing in healthcare and law.