Removing Silos to Combat Fraud, Waste in Veteran Benefits

Removing Silos to Combat Fraud, Waste in Veteran Benefits

The agency has encouraged cross-departmental collaboration and adoption of best practices to combat misuse of benefits.

The Department of Veterans Affairs is using cross-departmental collaboration and the application of new technologies to combat fraud, waste, and abuse within its benefits system.

Much of this centers on the burgeoning use of robotic process automation (RPA) capacities and machine learning to assist evaluators in more quickly and accurately detecting instances of benefits misuse. Speaking at the GDIT How AI can Prevent and Detect Fraud, Waste and Abuse forum, VA’s Program Manager of Fraud Prevention Services Erick Zenteno detailed how new practices have allowed the agency to overall improve the management of its benefits system.

As an initial priority, the Veterans Benefits Administration has focused on consolidating its business lines to allow for a more unified approach to fraud prevention.

“One area that was a challenge for us was fragmentation. Operations were siloed across all the Veterans Benefits Administration business lines. So that really was an issue in terms of integrating and centralizing efforts at an enterprise level. In some areas, some of the solutions that we had were good at that time, but were not best in class or did not have the benefit of commercial best practices,” Zenteno said.

As a means of improving VA’s broader waste and fraud prevention efforts, agency technologists focused on data consolidation as a means of more effectively evaluating potential instances of misuse through a unified process. This digital transformation program also involved using methods with a proven history of efficacy across the private sector.

“We had some challenges regarding data access and data sets that we needed to implement solutions for," Zenteno said. "The solutions that we put in place can be categorized into four different buckets — the first one being evaluating commercial best practices, second one being technology, the third being data resourcing, and the last one being a holistic approach in terms of integrating these commercial best practices."

VA’s adoption of practices used by major financial institutions seems to have been particularly instrumental in modernizing its fraud and waste detection program, with the VA having created a model for integrating these methods while sharing proven methods with other federal agencies as well.

“We've been able to establish an advisory council. This advisory council is made up of leading financial institutions with the goal to share best practices, insights, and what the banks are doing to mitigate and prevent fraud. Those are some of the lessons that we've been able to implement in terms of technology. Currently, they're working on a new model that will be able to give us different, more meaningful results in terms of fraud incidents that we encounter. We've been able to strengthen partnerships with not just internal stakeholders, but also external stakeholders, other federal agencies, to be able to share and ultimately try to be on the same page,” Zenteno said.

The ultimate outcome has been a facilitated process of using advanced data analytics and machine learning to make fraud and waste detection more accurate — as well as improve the VA’s ability to distinguish human error from intentional benefits misuse. This has allowed VA’s adjudication process to be easier to navigate for both civil servants as well as veterans and veteran families who might make innocuous errors.

“It’s not all criminal," Zenteno said. "Human error can be categorized rightly as unintentional errors ... sometimes the systems are too complex to navigate and errors happen. So part of the emphasis is to identify that and properly label it in a way that you can address. If something has human error, it doesn't do us any good to categorize that as fraud because there's not really fraud, it is just something that can be addressed by different ways."

 
Standard