Case Study: Global Insurance Company achieves scalable, automated workers' compensation fraud detection with H2O.ai

Global Insurance Company - Customer Case Study

Global Insurance Company needed to detect and prevent claims fraud in its Workman’s Compensation business—a problem industry-wide estimated at $80 billion annually. The company had consolidated large, mixed-format datasets (including handwritten medical notes) into Hadoop, but relied on manual examiner review and R-based analytics that could not scale to Hadoop volumes or avoid time‑consuming data extraction. H2O.ai was engaged to provide scalable, in-cluster analytics using H2O so analysts could work with Hadoop data directly.

H2O.ai deployed H2O co‑located in the customer’s Hadoop cluster, enabling data scientists to use their R workflows while running models at Hadoop scale without extracting or sampling data. Models are exported as Plain Old Java Objects (POJOs) for rapid, organization-wide deployment in Java-based transaction and case-management systems. The H2O.ai solution made fraud analytics more agile, reduced the backlog and data‑preparation time, and accelerated model deployment and scoring across production systems.