If anyone has experience in cybersecurity, it’s credit card companies. And, not surprisingly, AI has become one of their latest weapons in the struggle against financial fraud.
“We securely store over 18 petabytes of sensitive data — this is a significant amount of data that we’re dealing with every day,” Beth Griffin, vice president-Healthcare, Cyber & Intelligence, Mastercard, noted in a recent interview. “And we detect and defeat over 200 attacks on our network every minute of every day.”
Given that experience, it wasn’t a big leap a couple years back when the company decided to work to bring its experience to bear on the healthcare sector in the name of tackling the fraud, waste and abuse (FWA) that costs the sector an estimated $240 billion per year.
Mastercard, of course, has been deeply involved in the payment side of healthcare for 20 years, but according to Griffin it was “when the company started looking more closely at the healthcare system, it realized it had relevant capability when it came to using artificial intelligence (AI) to secure the system in such a way that the (payment) process moves smoothly for legitimate interactions while locking out the bad ones.”
The core problem, explained Griffin, is that healthcare can be an easy target for fraud because there’s an abundance of weak points. “From a transactional perspective, what typically happens today is that a claim file is sent from a provider to the insurance company payer who edits and adjudicates the claims, and then sends the file and the payment back to the provider.”
Only after that transactional process has run its course does the claim get reviewed by investigators looking for fraud. “The problem is that when they identify something that may be fraudulent, wasteful or inappropriate, they have to go back and try to get reimbursement from the provider. That often doesn’t go all that well — only an estimated 5 percent to 10 percent of FWA funds are ever recovered.”
So what does all this have to do with AI?
According to Griffin, Mastercard proposes tapping AI to move the fraud fight efforts up in the process, so that fraud is identified before it happens, and funds are blocked before overpayments are made.
The goal isn’t to eliminate the existing processes or the investigators that work with them, but, by building AI into the process, to make their jobs easier and more efficient.
“We’re not trying to replace what exists today — we’re really trying to complement what exists today,” she said. “In the end, if I’m in a special investigative unit, I can take advantage of the AI on top of what we’re already doing, and it decreases the false positives. So instead of having 500 alerts per day to investigate, my team might have just 50 that are highly likely to be fraud or abuse of the system."
In 2017, the company acquired AI firm Brighterion, and according to Griffin when they apply the Brighterion AI to the transaction flow in legacy solutions, they see a 10-20 times reduction in false positives. And they are spotting on average 2-3 times more fraud that has managed to slip under the radar, in some cases for years.
AI isn’t going to eliminate FWA overnight, but as Griffin sees it, since an estimated 3 percent to 10 percent of all healthcare costs are attributed to FWA, healthcare sector leaders are more open than ever to the idea that now is time, with new AI, to begin to do something about it.