Finding Fraud in Health Insurance Data with Two-Layer Outlier Detection Approach

Conventional techniques for detecting outliers address the problem of finding isolated observations that significantly differ from other observations that are stored in a database. For example, in the context of health insurance, one might be interested in finding unusual claims concerning prescribed medicines. Each claim record may contain information on the prescribed drug (its code), volume (e.g., the number of pills and their weight), dosing and the price. Finding outliers in such data can be used for identifying fraud. However, when searching for fraud, it is more important to analyse data not on the level of single records, but on the level of single patients, pharmacies or GP's. In this paper we present a novel approach for finding outliers in such hierarchical data. Our method uses standard techniques for measuring outlierness of single records and then aggregates these measurements to detect outliers in entities that are higher in the hierarchy. We applied this method to a set of about 40 million records from a health insurance company to identify suspicious pharmacies. © 2011 Springer-Verlag