AI could improve the necessary system of checks and balances between the provider and payer of health care whenever disagreements around the medical necessity of coverage arise.
In a recently filed class-action lawsuit, the families of two deceased patients contend their loved ones were denied coverage on the basis of an AI algorithm. That algorithm, in this case, happened to be proprietary to the patient’s medical insurance provider.
Regardless of the merits of the lawsuit, the headlines around the case suggest there is inherent bias in the application of AI, tipping the scales in favor of the business interests of one party or the other. While AI in itself is unbiased, the biases it exhibits are a reflection of the data and decision-making criteria established by its creators. Suggesting otherwise ignores AI’s tremendous potential to improve the cumbersome process of determining what constitutes “appropriate” medical services for patients.
According to an October 2021 report by McKinsey & Co., of the $4 trillion spent annually on health care, nearly 25% is spent on activities outside of the delivery of patient care. Health care spends more on administrative functions than any other U.S. industry. Further, it is estimated that $265 billion of the expenditure on these functions — including negotiations between doctors and insurance providers — could be eliminated without any negative impact to the quality or accessibility of care.
Use of an AI algorithm to remove the friction, error, waste and delay presents an opportunity for industry disruption — specifically, by improving the necessary system of checks and balances between the provider and payer of health care whenever disagreements around the medical necessity of coverage arise.
Applying an AI algorithm to determine the appropriateness of services is, in a way, the easy part. The digitization of medical records allows AI tools to read and interpret a person’s current medical profile against the backdrop of an entire population. Further, AI can objectively analyze medical necessity based on past examples where provider and payer agree on care decisions. The entire process – an inefficient task for even the best human data-crunchers – can be sped up by hours or days, if not weeks, through AI.
Ensuring the fairness of such an algorithm is the hard part. Efforts to train AI to reduce the trillion-dollar administrative burden on the health care industry are noble, but any algorithm must stand “in the middle” as a neutral third party, bridging the gap between payers and providers.
Not only can such an algorithm hasten the time needed to decide approvals, denials, and appeals, it could help turn a contentious process between payers and providers into one of collaboration. By spending less time on these kinds of administrative tasks, the health care system can focus its increasingly precious resources where they were intended: increasing the accessibility and quality of patient care delivery.
Joan Butters is a co-founder and the CEO of Xsolis, the AI-driven health technology company with a human-centered approach. Joan co-founded Xsolis in 2013 to help shape intelligent decision-making for payers and providers.