Application of clustering methods to health insurance fraud detection

Health insurance fraud detection is an important and challenging task. Traditionally, insurance companies use human inspections and heuristic rules to detect fraud. As the size of databases increases, the traditional approaches may miss a great portion of fraud for two main reasons. First, it is impossible to detect all health care fraud by manual inspection over large databases. Second, new types of health care fraud emerge constantly. SQL operations based on heuristic rules cannot identify those new emerging fraud schemes. Such a situation demands more sophisticated analytical methods and techniques that are capable of detecting fraud activities from large databases. The goal of this paper is to understand and detect suspicious health care frauds from large databases using clustering technique. Specifically, this paper applies two clustering methods, SAS EM and CLUTO, to a large real-life health insurance dataset and compares the performances of these two methods.