An Anomaly Detection Method for Medicare Fraud Detection

With the improvement of medical insurance system, the coverage of medicare increases a lot. However, while the expenditure of this system is continuously rising, medicare fraud is causing huge losses for this system. Traditional medicare fraud detection greatly depends on the experience of domain experts, which is not accurate enough and costs much time and labor.In this study, we propose a medicare fraud detection framework based on the technology of anomaly detection. Our method consists of two parts. First part is a spatial density based algorithm, called improved local outlier factor (imLOF), which is more applicable than simple local outlier factor in medical insurance data. Second part is robust regression to depict the linear dependence between variables. Some experiments are conducted on real world data to measure the efficiency of our method.