Medical Fraud Detection

Case Description

A state Medicaid agency covering about one million people throughout the state wanted to identify and reduce the number of fraudulent medical transactions. To accomplish the task of fraud detection, the agency hired us. The Medicaid agency wanted to identify and flag suspicious providers and collections of transactions, which could indicate fraud and thus required further investigation.

Implemented Solution

In order to reduce the volume of data to be analyzed and concentrate on more probable candidates for fraud, our analysts decided to first isolate and study records of patients who received more than 150 procedures during one calendar year. The resulting data contained about 400,000 transactions, which were further explored with the help of advanced analytical algorithms.

The Summary Statistics algorithm demonstrated that the remaining records corresponded to 998 patients, 623 types of performed procedures, and 903 providers, 24 of which were larger hospitals performing more than 5,000 individual transactions annually.

Provider-Patient Fraud

In order to identify possible provider-patient frauds, such as waiving patient co-pays or deductibles and over-billing the insurance carrier or benefit plan, the data was first investigated. Our analysis reveals and visually displays correlations between values of different attributes: the heavier is the line representing a link, the more correlated are the objects connected by this link. A more detailed analysis displays additional patient-provider pairs in a more organized fashion and allows the user to easily drill through to the underlying transactions and verify the validity of individual transactions.

Ghost Patient Billing

We uncovered groups of patients appearing in transactions performed by several providers, and identified groups of providers rending services to the same patients. The results revealed several groups of providers sharing a large number of patients. One particular group of three providers was sharing over 10% of all patients consider in our analysis. There is always a change that these providers offer complementary services and the fund transactions are perfectly legitimate.