Big data and AI are powerful tools in the fight against health care fraud - SmartBrief

All Articles Healthcare Big data and AI are powerful tools in the fight against health care fraud

Sponsored

Big data and AI are powerful tools in the fight against health care fraud

Fraud, waste and abuse is a costly problem in health care, but data, analytics and AI could turn the tide.

4 min read

Healthcare

Use big data to fight fraud

Shutterstock

The National Healthcare Anti-Fraud Association estimates that health care fraud costs tens of billions of dollars every year. Health insurance special investigations units have long worked to stem the tide of losses, but it can be a challenge to stay ahead of the curve. To learn more about the tactics investigators employ and new tools at their disposal, SmartBrief spoke with Patrick Stamm, principal adviser to FraudScope, an AI-assisted platform for detecting health care fraud, waste and abuse.

Please describe the existing health care FWA investigation paradigm and some ways in which it fails health plans?

Patrick Stamm

Historically, health care fraud investigations were almost exclusively initiated after the fact — often as a result of a tip submitted to the special investigations unit from a member or employee who noticed something amiss while processing a claim. So SIUs depended on people noticing something wrong and then being confident enough to report it. As we started to see analytics being used more often, the first place this technology was applied was to large amounts of claims data. These were historical paid claims that enabled investigators to apply rules, queries or analysis to pick up outliers or other suspicious characteristics and flag providers for investigation. 

Both of these scenarios were retrospective by design and quite costly and time-consuming. Another challenge has been that after a fraudulent claim is paid, it becomes very difficult to get all the money back.

What are some of the key data streams SIUs should stay on top of, and how can technology help them make sense of it all?

Certainly at the top of the list are claims data, provider history and profiles, and member eligibility information, which most SIUs have access to and are using. Adding data for things like preauthorizations, laboratory requests, prescriptions and pharmacy, and even call-center records will enable analysis. Obviously, with big data technologies maturing and becoming mainstream, our ability to access, store and use all this data has improved and is become more real-time. 

FWA investigators have long been challenged to spot emerging types of fraud. Why is it so difficult to stay ahead of the curve?

I think most SIUs do a pretty good job of spotting fraud — even emerging types — when they see it.  One of the biggest challenges is being able to flag suspicious claims or providers and get them to the SIU so they have the opportunity to investigate. Think about it: Most health plans will see many millions of claims each month. The capability most needed is to be able to sort through those claims and find the few that need to be reviewed and investigated. The use of big data technologies and advanced analytic techniques has made this much easier, but we are in the early stages of widespread deployment.

How can artificial intelligence improve FWA detection?

I think the combination of big data and artificial intelligence is a very significant opportunity in our antifraud efforts. With big data, we now have access to so much more information and on a much more real-time basis. Applying AI to that data gives us the ability to use machines to not only create more advanced and accurate analytics, but also to do so in a much faster way. This does a couple of things for us: First, by using risk scoring versus more traditional rules-based analytics, we can cast our net wider but at the same time get a more accurate analytic to help reduce false positives and help reduce abrasion in the provider community. Second, the increased speed allows us to more easily conduct the evaluation of suspect claims before payments are made. This alone can improve recoveries by 15% to 20%.

Why are workflow and collaboration tools important to allow SIUs to work more efficiently and effectively?

SIUs rely heavily on experts across the organization to assist in gathering facts and providing evaluations as they investigate suspicious claims. Emerging collaboration tools make that coordination much easier and more efficient, and these tools can help reduce the time it takes to effectively complete investigations and finish cases. This makes SIUs and their critical and skilled resources more productive at a time when most organizations are facing increased levels of fraud.

Patrick Stamm works to enable organizations to improve through the application of technology, analytics and artificial intelligence. He serves as principal adviser to FraudScope and previously held the role of chief operating officer for shared service operations at UnitedHealthcare, where he built and developed a variety of teams and capabilities focused on big data and advanced analytics; ML and AI; and tools and techniques to identify errors and FWA, eliminating more than $10 billion in medical cost losses per year.