“Chasm crossing is not the end, but rather the beginning of mainstream market development.”
Geoffrey A. Moore in Crossing the Chasm
Last week, we started our discussion of just how AI in the form of robotic process automation and machine learning can help reduce the burden of one of the challenging aspects of healthcare administration, the dreaded revenue cycle management. We can elaborate on just how ML and RPA can help most if not all aspects of the entire RCM process by detailing how AI can help two of the most demanding tasks:
1. Prior authorizations
This tedious task under claims preparation involves obtaining approval from the payer for a diagnostic test or therapeutic procedure prior to the execution of the procedure. For the clinicians, this task is more than the inconvenience of the time spent on this process but more importantly its delay that can lead to dire consequences for their patients.
Artificial intelligence in the form of RPA and machine learning can facilitate this convoluted process by creating an effective interface that will lead to higher probability of real-time approval at the time of physician order entry. This higher approval rate is achieved by scrutiny of past denials and approvals and making recommendations during the time of order entry to neutralize the elements that will lead to denials. This strategy can engender a portfolio of payer and procedure-specific prior authorization templates that will lead to reduction of turnaround times and increase in denial reconciliations, and ultimately increase in revenue.
2. Claims denials
This process under claims management of the RCM can be categorized into “soft” denials that are temporary and simply require additional documents or “hard” denials that do require appeals. Both types of denials can result in significant loss of revenue as some hospitals do not routinely appeal the denials even if denials of first submissions are not very high (<10%).
Artificial intelligence can be leveraged to decrease the number of claim denials to as low as possible by proactively identifying and avoiding “at-risk” scenarios that lead to denials prior to claims submission. The strategies can include identifying the payers and CPT codes that result in denials, providing missing information or correcting incorrect information, streamlining the process to avoid being late with deadlines, and focusing on claims with higher values. As mentioned under prior authorizations, a growing number of templates of payer and procedure-specific prior authorization templates can also be used to decrease claims denials.
In addition to the aforementioned processes, RPA can help with tasks such as gathering and combining billing data from disparate sources; account reconciliations on a weekly or monthly basis; follow-up with payors; and cash posting. In essence, RPA as well as ML can help with any part of the RCM cycle.
The vision of an AI-enabled, real-time revenue cycle management for a health system may be just a few years away, but adoption of AI for RCM in most hospitals remains less than robust for now. We are indeed in the “early adopters” era of AI in RCM and hospital financial management, and facing a “chasm”.
