Foretelling the physician’s future: Health care fraud prediction with behavioral big data

Fraud, waste, and abuse have been identified as some of health care’s most vexing challenges, contributing to rising health care costs.

The cost of physician fraud is estimated at hundreds of billions of dollars or 10 percent of all medical spending, borne by federal and state government, patients, and taxpayers, there is a pressing need for early prediction so that negative consequences can be averted.

In a project funded by the U.S. Department of Justice, CDHAI researchers worked in collaboration with faculty from the criminology department at the University of Maryland to develop a comprehensive dataset combining behavioral “big data” on physicians, including purchasing patterns, background information, and online reputation, with federal and state criminal records. This dataset was utilized to construct a matched set of fraudulent and non-fraudulent physicians. Machine learning was used to develop a predictive model with high accuracy that predicted fraudulent behavior with a one to five year look-ahead.

This model, once implemented, can reduce health care fraud via early education behavioral interventions. It can also be applied to improve fraud detection accuracy.