Blogs

Post a Comment

Big Data & Analytics Helps Government Agency Identify Improper Payments

October 14, 2013

When weeks of analysis failed to uncover improper payments, the Big Data Analytics engine did it in 4 hours – and led to a $140 million payoff

Needs/Challenges:

This large government agency provides both medical and social benefits to a very large group of beneficiaries.  It is responsible for ensuring its programs and operations are efficiently and effectively managed.  It must also prevent and detect criminal activity, waste, abuse and mismanagement. The agency was having difficulties in this area: analyses to identify criminal activity took weeks, and were still incomplete.  The analyses required ad-hoc data requests from 70 separate data sources: in-patient, out-patient, prescriptions, financial records, notices of death, criminal data, many others. They needed an agile environment to speed up analyses, identification, prevention and prosecution. 

Solution:

The IBM Big Data Analytics engine was brought in and after the data was quickly loaded, it was able to reduce from weeks to hours the time it takes to run analyses that helps pinpoint criminal activity. The flexibility and time savings that the system offers had significant, positive impact when accessing information that requires immediate attention and impacts the safety and well-being of the beneficiaries. The system also has simplified analytic operations, delivering results much faster and significantly reducing its operating costs.

Benefits:

  • The Agency was able to identify improper payments of over $140 million dollars, including benefits to persons who had already died.
  • Contributed to a 35-fold improvement in savings and cost avoidance between 2005 and 2012 
  • Critical analyses are able to be done in hours instead of weeks, improving agency, efficiency, effectiveness and saving taxpayer dollars
  • Minimal DBA requirements
  • Analytics to identify criminal activity takes minutes instead of weeks. 
  • Enables ad-hoc analysis of 70 separate data sources
  • Return on Investment was achieved in 18 months