Intelligent Automation of the Revenue Cycle: How an Integrated Platform Approach Yields Financial Results

Sean Barrett, Senior Vice President, Product and Digital Transformation, R1 RCMHealthcare provider networks are experiencing enormous pressure to manage financial margins and invest in contactless patient experiences. With overall financial losses projected to exceed $323 billion as a result of COVID-19, a projected $200 billion in administrative waste due to revenue cycle inefficiencies, and increasing pressure to meet digital consumerism demands, it is essential for health systems to find ways to streamline processes, maximize their revenue cycles and cut costs. These industry trends are pushing organizations to invest heavily in automation solutions, such as artificial intelligence (AI) and robotic process automation (RPA) to alleviate operational and financial pressures. 

In this rush to invest in automation and digital solutions, providers are often overlooking how a multi-layered technology approach can increase value realization. They need an intelligent automation (IA) platform that incorporates a mixture of powerful AI technology levers such as machine learning (ML), natural language processing (NLP), and optical character recognition (OCR), combined with RPA and workflow orchestration, which enables humans to work harmoniously with these digital assets. This article will examine how this IA platform can be utilized to strategically deliver financial value within the revenue cycle. 

When deployed correctly, IA can help health systems realize new revenue streams by improving net revenue capture, deliver cost reductions through automating time-consuming rules-based revenue cycle tasks and produce more predictable reimbursements. However, in order to achieve these financial and operational results, organizations need to assess how and where to apply technology.

If working with unstructured data such as an image file or clinical chart, NLP and/or OCR technology needs to be deployed to pre-process or extract data; however, if you are working with large volumes of structured data, ML can be utilized straight away to assess trends and determine the best way to complete a transaction. When completing repetitive and routine revenue cycle transactions, such as adjustments, insurance verifications, and payment postings, RPA may be the right choice since it employs digital workers to perform these actions accurately and quickly. 

While these technology levers deliver major enhancements individually, when utilized together they act as multipliers – expanding the number of revenue cycle challenges that can be solved through automation. With an IA platform, health systems can address several distinctive issues all while continually removing waste from revenue cycle processes and creating more capacity operationally. 

With NLP and OCR technology, organizations can convert unstructured data from files that are frequently utilized in healthcare – medical records scanned documents, and audio recordings – into structured, normalized data.  For example, OCR can convert explanations of benefits (EOB) PDFs into a data table that RPA bots can then auto-post into patient records. NLP can extract clinical terms from an EMR note and provide key data elements to a machine learning model that will then assess the likelihood of medical necessity denials prior to adjunction. In these scenarios, NLP and OCR are translating everyday documents into workable data for faster processing and applying the full IA platform to generate cost optimization and improve revenue capture across a health system’s enterprise. 

An IA delivery platform also gives health systems access to better decision-making tools since the technology can consume large volumes of data and subsequently create learning algorithms that make consistent decisions on behalf of operators based on the task at hand. For example, ML can apply historical claim reimbursement trends when assessing data to predict potential write-offs and then using the integrated workflow platform to either escalate high priority items to operators or direct low-dollar write-offs to RPA to process. These learning algorithms can be applied to many situations within the revenue cycle to achieve greater cost optimization and streamline revenue cycle operations.  

Finally, while most revenue cycle processes can be fully automated, there are still exceptions and use cases that require human intervention. Automation technology should be paired with an integrated workflow platform that can determine if a revenue cycle task should be automated or handled by humans to create a natural orchestration and seamless hand-off between digital workers and humans. 

To illustrate how the IA platform truly works, let’s look at a common workflow: correspondence management. Paper document processing is still highly prevalent in healthcare, requiring significant resources from health systems. For example, issuing correct billing correspondence to patients requires receiving paper correspondence from banks, such as letters, checks, and EOBs, reviewing what is typically tens of thousands of files per day, and manually entering data from these files into subsequent workflow solutions.

With an IA delivery platform, this process can be automated by utilizing RPA to retrieve these documents, OCR and NLP technology to convert these documents into standard file formats, and then, once again, RPA to process and attach necessary documents into patients’ accounts in the accounting or indexing system. While these activities are taking place, the integrated workflow platform is tracking the activity and flagging any exceptions or high-risk materials that need to be pulled out and handled by humans. 

This symbiotic platform creates standardized processes patients can rely on and that can ultimately be scaled. Patient experiences are improved by reducing frustrating administrative errors, such as misplaced information, incorrect bills or inefficient handoffs, that can prolong billing cycles. Health systems also typically see reduced revenue leakage and lower cycle times with automated processes since manual errors are significantly reduced and automation runs processes 24 hours a day.

Since an IA platform can take on numerous revenue cycle challenges across a health system’s enterprise and standardize them – removing many common administrative errors or interoperability issues – it gives leaders more visibility into daily operations allowing them to be more proactive in finding opportunities to boost revenue streams (e.g., ways to eliminate revenue leakage, increase ease of scheduling/payment for patients or maximize patient volume) and continually improve performance. Thoughtful application of these technologies in a platform-based strategy enables provider organizations to improve revenue and reduce costs so that they can prioritize their mission of keeping their patient population healthy.  

Now is the time to move past a “one technology fits all” mentality and find solutions that can strengthen and improve multiple revenue cycle workflows and tasks. Given the high cost associated with developing these digital capabilities, health systems need a partner who not only offers the right model, but has made necessary investments toward cutting edge IA research, fully-staffed teams with subject matter expertise, and data-rich analytics needed to foster ongoing performance improvement. With an aligned partnership and IA platform, health systems can produce successful results that achieve intended benefits.

About Sean BarrettSean Barrett is the Senior Vice President of Digital Transformation at R1 RCM. He joined R1 in 2018 and currently oversees R1’s core product management, automation, and machine learning functions. Prior to R1, Sean spent 14 years at Deloitte Consulting focusing on serving clients primarily in the healthcare provider segment-leading operational performance improvement and technology-driven transformation engagements at many of the largest health systems in the country.

Show CommentsClose Comments

Leave a comment

%d bloggers like this: