A Novel Method for Fraudulent Medicare Claims Detection from Expected Payment Deviations

Healthcare has and continues to be an integral component in people's lives, especially for the rising elderly population. One such healthcare program that provides for the needs of the elderly is Medicare. It is important that any such program be affordable but, unfortunately, this is not always the case. Out of the many possible factors for the rising cost of healthcare, fraud is a major contributor, but its impacts can be lessened through the use of fraud detection methods. We assess possible fraudulent activities by looking at the amounts paid to providers for services rendered to patients. In this study, we propose a novel methodology and framework towards identifying potential sources of fraud. We model these Medicare payments in order to create baseline values that reflect what the payments should be for a provider's specialty. We use these baseline expected payments and compare them to what was actually paid by Medicare for distinct specialties and healthcare services. Any deviations from the expected payments are flagged for further investigation. Our overall approach is consistent with related works, in healthcare, using anomaly-based detection methods to detect fraudulent activities, but we focus on an implementable and generalizable framework for initial fraud detection. Our results demonstrate the detection of possible fraudulent activities, with one specialty, Cardiology, demonstrating the detection of a known, real-world fraud case.