Detecting Health Insurance Claim Fraud Using a Mixture of Clinical Concepts

Abstract

In order to pay for the expensive medical treatment, patients rely on health insurance offered by either the public or private sectors, or both. Some medical professionals perpetrate insurance fraud because their patients rely on them. Even if there aren't many of them, insurance companies reportedly lose billions of dollars annually as a result of fraud. This work presents a formulation of the fraud detection issue over definite claim data consisting of medical operation and diagnostic codes. The data set is modest in size. By converting procedure and diagnostic codes into Mixtures of Clinical Codes (MCC), our innovative representation learning technique allows us to identify fraudulent claims. We further explore potential expansions of MCC
using Long Short Term Memory networks and Robust Principal Component Analysis. Our experimental results demonstrate promising outcomes in identifying fraudulent records.