Not all AI is created equally.  Artificial Intelligence insurance fraud solutions can vary greatly, directly impacting the accuracy of various healthcare fraud, waste, and abuse (FWA) solutions. “True” AI is defined as a system that exhibits behavior that is at least as accurate and as flexible as a human auditor.  These systems have also been described as having the ability to improve by utilizing Deep Learning or Reinforced Learning mechanisms.

Most traditional FWA software solutions try to integrate rudimentary AI into their FWA platforms for marketing and regulatory reasons but do not have true AI capability.  This creates a rigid, rules-based system with some AI, not a robust AI system that learns from rules and adjusts from human interaction. To be clear, rules are not bad.  In fact, they are an important part of the FWA answer.  But rules alone with a sprinkling of superficial AI is simply not enough to battle complex fraud schemes.

Anything less than a true AI system will produce the same mistakes and false positives over and over again, never adjusting and never producing different output to the user.  Basic systems will never improve to the point where a client can allow it to act on a claim in a prepay environment. And they can never have the flexibility that is crucial in a changing medical healthcare claims environment.

Today’s Corona-19 pandemic is a perfect example of how quickly rules and regulations can change.  Many policies and procedures that were not allowed by CMS and Medicare only a few short weeks ago are now allowed.  The industry is seeing pre-authorization and provider eligibility changes overnight. An effective FWA system must have the flexibility and adaptability to constantly adjust to the changing regulatory landscape with little or no human interaction. Systems that do not use true AI and cannot self-adjust can do little in a changing environment to identify a provider who is billing and operating differently from their peers or recognize abnormal behavior in the changing billing landscape.

Many prominent FWA software companies tout the use of thousands or even millions of rules built into their system. In a changing regulatory environment, growing and updating a rules-based system becomes stressful and expensive, requiring the manual construction of new rules daily. Just imagine changing and managing thousands or even millions of rigid rules in today’s fluid environment; it would take an army of staff. True AI systems are crucial now and in the future.

Leave a Reply