“Insanity is doing the same thing over and over and expecting different results.”
We have had frequent discussions about artificial intelligence and its impact in clinical medicine and healthcare, so it would also be good to focus on use of AI in non-clinical aspects in healthcare, such as administrative and financial tasks. Healthcare finance and administration challenges include staffing shortages, resource allocation, budget constraints, decreasing margins, and increasing denials; all of which have been impacted by the havoc from the pandemic.
I’d like to delineate the challenges of revenue cycle management (RCM), one of the most difficult sectors in healthcare administration, and how these challenges can be partly solved with machine learning (ML) and robotic process automation (RPA) tools in AI. Despite the high level of difficulty of RCM and the large sums of money involved in both the execution and dividend of RCM, the practices and strategies of this sector have not changed dramatically.
Revenue cycle management (RCM) involves claims by the providers and reimbursements by the payors, and this often leads to a push-pull from both sides. These two parties, however, would very much like to reduce the administrative burden and therefore decrease their administrative costs. The usual steps of RCM starts with 1) claims preparation with patient registration, prior authorization, patient eligibility and benefits verification, followed by 2) claims submission with charge entry, rules engine, and claims transmission, then 3) claims management with payment posting, claim denials, accounts receivable (A/R) follow-up, and patient collection, and finally 4) reporting and analytics. Despite the complicated list of tasks embedded in a convoluted network, there has not been much innovation nor automation in RCM and its related tasks.
The steps of revenue cycle management can be improved with AI solutions, especially in the forms of ML and RPA. The RPA market size is expected to reach 10 billion USD while that of ML is expected to go well over maybe 200 billion USD by the end of this decade. While ML can self-learn from data (“data-driven” or “data-based”) and does not require traditional programming, RPA or RPA “bots” automate processes that involve repetitive tasks (so therefore “process-driven” or “rule-based”). In other words, RPA bots are usually confined to preset rules whereas ML takes data to enable a prediction function.
These tools and other synergistic and valuable resources such as natural language processing are poised to be essential elements in the optimization of processes, resources, and costs in healthcare in this decade and beyond. RPA can increase expediency, promote efficiency, decrease labor, improve accuracy, and reduce cost. Enterprise scalability and institution automation are additional benefits.