Skip to main content

Customer Experience and Design

Big Data to attack Healthcare Fraud

According to a U.S. Senate Panel on Healthcare, healthcare fraud and abuse costs as much as $100 billion a year, or 10 percent of the $1 trillion spent on healthcare. In addition, these fraud costs are increasing at a rate of $3 billion a year. The traditional approach has been to pay claims and then investigate or data mining on the post-adjudicated claims. The enormity of the healthcare fraud and abuse problem demands that healthcare plans shift their focus to avoidance or pre-adjudication of potentially fraudulent claims. This massive shift will require more real-time data mining, identification of patterns of abuse and a broader view of healthcare claim processing than the individual claim itself.

How does this fraud and abuse take place? The most common problem is inaccurate or fraudulent billing schemes. It takes powerful data mining software with the ability to define behavior models to detect these common fraud and abuse schemes. With the shift from ICD-9 to ICD-10, there is increased risk that claims can be miscoded for services that don’t meet coding or necessity policies as well. Most importantly, risk management departments for health plans need the ability to holistically examine large scale patterns from healthcare vendors, providers or services provided. The challenge is the huge volume of claims processed and the need for examining the out of compliance or outliers to reduce analyst workloads.

How does pre-adjudication fraud detection differ from current pay and chase methods? Proactive pre-adjudication analysis for fraud, waste or abuse requires a combination of two major technologies – predictive modeling and Big Data processing. Predictive modeling scores claims or a series of claims against pre-built fraud detection models that use thousands of potential risk indicators and combine continuous machine learning to uncover new fraudulent schemes. Big data processing, like Hadoop, allows the examination of large data patterns for small abuses and blocking those schemes that use small over-charges, for example.

Pro-active pre-adjudication of claims to stop fraud, waste and abuse will require accurate analysis of a claim with only partial data in some cases – a real challenge for today’s IT systems. Without sophisticated fraud management software that includes the predictive modeling, risk scoring and scheme pattern detection, the losses will continue to stack up. It’s past time to upgrade from out dated batch processing of claims and move to near real-time pre-adjudication of claims using Big Data techniques and predictive modeling. The reward is huge and more than cost justified for the IT investment.


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Martin Sizemore

Enterprise Architect with specialized skills in Enterprise Application Integration (EAI) and Service Oriented Architecture (SOA). Consultant and a trusted advisor to Chief Executive Officers, COOs, CIOs and senior managers for global multi-national companies and healthcare organizations. Deep industry experience as a consultant in manufacturing, healthcare and financial services industries. Broad knowledge of IBM hardware and software offerings with numerous certifications and recognitions from IBM including On-Demand Computing and SOA Advisor. Experienced with Microsoft general software products and architecture, including Sharepoint and SQL Server. Deep technical skills in system integration, system and software selection, data architecture, data warehousing and infrastructure design including virtualization.

More from this Author

Follow Us