Abstract: 

According to FBI, total health spending in America was around $3.3 trillion in 2015 and spending continuing to outpace inflation. According to several recent studies, estimates for healthcare spending lost due to fraud, waste and abuse (FWA) ranged between $90B and $330B! At the same time, only the small portion of this money is recovered 2014 from individuals and companies who attempted to defraud various health care programs.

The talk will show how we are developing a comprehensive, preventive and intelligent analytical framework for detecting fraudulent, abusive and wasteful claims in near real time using big data technologies and advanced machine learning techniques. The speaker will provide a quick overview of our in-house FWA detection framework and demonstrate how it is used to discover specific FWA scenarios identified in the most recent months.

The presentation will explain why there is a need to choose efficient data infrastructure to handle ever-growing health care data and why we have chosen Hadoop environment to efficiently handle millions of claim data records on a daily basis. Participants will learn about various data sources (in addition to claim data) that are used to detect fraudulent doctors and facilities. In addition, the speaker will offer practical advices how to effectively organize multiple data sources, how to appropriately set the problem and how to design an effective predictive / data analytics solution that will not only detect suspicious FWA leads but will also direct investigators through their faster review by providing effective visualization and suggesting most likely reasons behind such leads.

Bio: 

Aleksandar Lazarevic, Ph.D., Senior Director at Aetna Data Science Organization Aleksandar is responsible for overall predictive analytics solution in health care fraud initiative at Aetna. In addition to health care industry, he has extensive experience in various analytics projects ranging from banking, credit and insurance industry to diagnostics and computer security applications. He has co-edited a book on cyber security threats, written 8 book chapters and published over 50 research articles, which were cited more than 3,500 times. He holds a PhD degree in data mining / machine learning from Temple University.