Using support vector machines to detect medical fraud and abuse

This paper examines the architecture and efficacy of Quash, an automated medical bill processing system capable of bill routing and abuse detection. Quash is designed to be used in conjunction with human auditors and a standard bill review software platform to provide a complete cost containment solution for medical claims. The primary contribution of Quash is to provide a real world speed up for medical fraud detection experts in their work. There will be a discussion of implementation details and preliminary experimental results. In this paper we are entirely focused on medical data and billing patterns that occur within the United States, though these results should be applicable to any financial transaction environment in which structured coding data can be mined.