DETECTION OF FRAUD CLAIMS IN HEALTH INSURANCE INDUSTRY

Abstract

For patients to pay for the expensive medical bills, they rely on health insurance offered by either private, public, or both systems. Some healthcare practitioners conduct insurance fraud as a result of their reliance on health insurance. Despite the tiny number of these service providers, it is said that fraud costs insurance companies billions of dollars annually. Our study involves formulating the fraud detection issue over minimum, definite claim data, which consists of procedure codes and medical diagnoses. Using a unique representation learning technique, we provide a solution to the fraudulent claim detection issue by converting procedure and diagnostic codes into Mixtures of Clinical Codes (MCC). We also look on ways to extend MCC using Robust Principal Component Analysis and Long Short Term Memory networks. Our test findings show encouraging results in the detection of false records.