A Medical Claim Fraud/Abuse Detection System based on Data Mining: A Case Study in Chile.

TLDR: The application of data mining to a real industrial problem through the implementation of an automatic fraud detection system changed the original non-standard medical claims checking process to a standardized process helping to fight against new, unusual and known fraudulent/abusive behaviors.

Abstract: This paper describes an effective medical claim fraud/abuse detection system based on data mining used by a Chilean private health insurance company. Fraud and abuse in medical claims have become a major concern within health insurance companies in Chile the last years due to the increasing losses in revenues. Processing medical claims is an exhausting manual task carried out by a few medical experts who have the responsibility of approving, modifying or rejecting the subsidies requested within a limited period from their reception. The proposed detection system uses one committee of multilayer perceptron neural networks (MLP) for each one of the entities involved in the fraud/abuse problem: medical claims, affiliates, medical professionals and employers. Results of the fraud detection system show a detection rate of approximately 75 fraudulent and abusive cases per month, making the detection 6.6 months earlier than without the system. The application of data mining to a real industrial problem through the implementation of an automatic fraud detection system changed the original non-standard medical claims checking process to a standardized process helping to fight against new, unusual and known fraudulent/abusive behaviors.