AI/ML

Making the case for using AI for healthcare fraud, waste and abuse detection


In a webinar sponsored by AHIP, we demonstrated the value of using AI for healthcare fraud, waste and abuse prevention. Presented by leading claims investigators, the case for AI has never been more relevant than during a global COVID-19 pandemic.

Brighterion and Mastercard recently presented a webinar that discussed aspects of using AI for healthcare fraud and demonstrated AI’s superiority over rules-based systems. Though more commonly used, rules-based solutions have their limits. Here’s a summary of what Beth Griffin, Vice President, Healthcare Product Development and Innovation, Mastercard, and Tim McBride, Director, Healthcare Product Development and Innovation, Mastercard, had to say. The session was moderated by Jala Attia, President of Integrity Advantage.

As we’ve discussed in past posts, fraud, waste and abuse (FWA) is estimated to be at least $240 billion annually in the U.S. Jala believes it’s even higher, and the Center for Disease Control expects that to climb with fraudulent COVID-19 claims. Payers are already overwhelmed by false alerts and investigations. How can they cope with increased volume?

A recent study by PYMNTS, The unlocking AI playbook: healthcare edition, found that while payers have limited resources and are concerned about FWA, they are skeptical of new technology. As Beth pointed out, payers use rules-based solutions and algorithms, currently but these are quickly outdated and require frequent updates.

How does AI stack up against current rules for FWA?

Tim offered a brief comparison of today’s FWA solutions with AI-based platforms. Rules-based systems are hard-coded and feature many algorithms. These rules are self-limiting and lose value over time as fraudulent behaviors change. The research, analysis and work to update the rules are expensive and time-consuming. Forty percent of the claims investigated are found to be false positives.

Compare that to using a single advanced AI model that provides its own updates through real-time, continuous self-learning, adapting to the changing behavior it observes. It also learns through the final disposition of cases and claims, so can identify valid claims and flag new behavior patterns that may indicate fraud. The end result is a reduction of 20x in false positives, providing likely-fraud alerts to investigators for follow up.

Tim also broke AI down to its most basic principles. AI is when computers collaborate and learn from each other, not unlike humans in a work group, to:

  • Achieve a goal
  • Use multiple machine-learning tools
  • Trigger actions and reactions
  • Make decisions or recommendations

Focus on real fraud and perform strategic work

As a result, special investigations units (SIUs) are much more efficient and successful. Fewer false positives allowed investigators to focus on real fraud and perform strategic work. During the Q&A session, one attendee asked if that meant SIUs would be eliminated. “Not at all,” Beth said. “It means they can be more efficient. Investigators can focus on very suspicious activities and stop payments from going out.

In an industry where claims are paid, then investigated – hence the phrase “pay-and-chase” – this is significant. Beth prefers a pay-and-save model, one that sees the AI flagging claims upon submission, triggering immediate action before payment.

FWA case studies and types of fraud 

A few case studies were presented, including one that showed increased accuracy by 38 percent (from 60 to 83 percent) in an internal medicine department, and found $1 million of previously unidentified fraud. The model was trained on the payer’s previous 18 months of claim data, with the rationale divided into two segments: what FWA behavior does and does not look like.

The model was built to the payer’s stated precision rate. As much as $1.5 million FWA was identified, but the payer chose to go with a lower precision rate to allow for provider errors and avoid provider abrasion.

So what are some of the things Mastercard’s AI can identify? It’s currently focused on four areas: claim, member, provider and pharmacy fraud. These types of fraud are further explored in our Ebook, Prevent and save: advanced AI for fraud, waste and abuse.

Tim discussed another case study, in which a provider grossly overcharged for services not provided to an advanced-stage cancer patient. That one case alone included $4,800 of genetic and molecular pathology tests for a person who was already receiving end-of-life care.

The potential for increased fraud during COVID-19

Which brings us back to the current pandemic. To enhance current patterns and codes, Mastercard’s data scientists are building models based on other outbreaks, such as H1N1, SARS and MERSA. The new models will be used to identify fraud linked to COVID-19 diagnosis and treatment.

They are also meeting telehealth providers to learn what restrictions have been lifted and to understand normal and fraudulent patterns of behavior in this service. Tim’s personal belief is that a lot of fraud will be seen via telehealth claims. He added that access to public records also allows the system to screen for provider sanctions, amongst others.

An opportunity to test drive AI for healthcare FWA

Mastercard has over 15 years of experience with fraud mitigation. We’ve developed a unique proof of concept process, AI Express, that builds a model based on an organization’s current data.

AI Express is a collaborative effort, mainly performed by a data science team, using Mastercard’s technology tools. Our process reduces the impact on our customers as we enrich the data and build the model for their specific goal. Customers have three opportunities to evaluate along the way. The model is built in six to eight weeks and the customer can test it to decide if they wish to deploy. It’s an opportunity to test drive AI for healthcare FWA. It’s an opportunity to test drive using AI for healthcare fraud.

The Mastercard Healthcare Solutions team comprises data scientists with healthcare and payment fraud experience, and investigators with extensive claims processing experience. An overall project manager provides oversight throughout.

Learn more about Mastercard Health Services and the opportunity to use AI to prevent fraud, waste and abuse, reduce false negatives, and gain higher efficiencies for your investigative staff.

The post Making the case for using AI for healthcare fraud, waste and abuse detection appeared first on Brighterion.

Source link

Related posts

Providers Turn to EHRs for Help in Tackling Measles Outbreak

Newsemia

Chatbot Trends You Need to Know Before Investing

Newsemia

Future of Housing: 3D printed neighborhood to come up in rural Mexico

Newsemia

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy

COVID-19

COVID-19 (Coronavirus) is a new illness that is having a major effect on all businesses globally LIVE COVID-19 STATISTICS FOR World