Philosophy & Research

---

Claims Sciences is an analytics platform centered around the detection of fraud, waste, and abuse (FWA) across public health programs, with a focus on provider-level utilization and billing behavior. The platform grew out of a series of academic research papers studying novel techniques in public healthcare fraud detection. Today, those academic roots drive our ongoing focus on identifying statistically abnormal patterns that signal emerging risk within large-scale claims data.

Providers who ultimately face enforcement or exclusion consistently exhibit detectable anomalies in their billing behavior months before formal action. By combining anomaly detection, peer benchmarking, and cross-dataset linkage (e.g., regulatory exclusion lists), our approach surfaces these providers significantly earlier than traditional methods. We uncover these patterns using the principle of parsimony; we believe that when data is viewed through the correct lens, anomalous behavior naturally stands out. To maintain this parsimony, we reject black-box models entirely, relying instead on strict explainability and the philosophy that program integrity is fundamentally a data-viewing issue.

This enables program integrity teams to move beyond retrospective 'pay-and-chase' models toward early interception. Our platform prioritizes high-risk outliers and generates actionable leads before improper payments compound.