Application of The k-means Clustering Algorithm In Medical Claims Fraud / Abuse Detection

Abstract

There is a serious threat to the survival of the insurance industry in Kenya due to fraudulent medical insurance claims. Several studies have shown that unsuspecting insurance companies lose huge sums of money annually settling fraudulent claims. It is feared that if this issue is not stemmed, many insurance companies offering the medical covers may collapse. The aim of this thesis was to determine the extent of fraudulent medical claims within insurance companies in Kenya, establishing the measures in use to counter the problem and thereby develop a system to detect these fraudulent activities. The systems was developed applying the k-means algorithm using mySQL and Java software as development tools. The k-means algorithm is well known for its efficiency in clustering large data sets. A major limitation of this algorithm is that it works only with numeric values, thus the method cannot be used to cluster real world data containing categorical values. However, this limitation was addressed by converting the data sets to numeric data whereby ailments were listed and matched with patients. The presence of the ailment was represented by a one (1) and the absence was represented by a zero (0).