Health Insurance Fraud Detection

Fraud & Risk

Prevent, detect and manage fraud across the enterprise, making smarter decisions, increasing return on capital and driving business performance

Health Insurance Fraud Detection

Using Big Data to Detect and Prevent Health Insurance Fraud

Fraud costs. A lot. According to the National Health Care Anti-Fraud Association health care fraud costs the country an estimated $68 billion annually (3% of the $2.26 trillion in health care spending). At a time when health insurance plans have a national mandate to reduce costs, the reduction and elimination of fraud is of paramount importance. Experfy can help your firm to use Big Data to address this pressing problem. 

Challenges and Opportunities

High health insurance fraud rates—the percentage of health insurance claims that are illegitimately processed—are a serious detriment not only to insurance companies but also to the broader society which ultimately pays the price in terms of higher fees in a nation wherein health care spending amounts to an astronomical 18% of GDP. Compounding the fraud problem is that the incredible amount of paper work in the health care field makes it an epicenter for a significant amount of the $17 billion in annual medical errors. Part of the specific challenge in addressing the fraud problem is that governmental regulations prevent insurers from delaying claims payments unless they have strong evidence that the claim is in fact fraudulent. Current methods have made that very difficult to do in real time so loss recovery has historically been conducted retroactively which makes the process far more complicated and far less likely to succeed. For example, identification is central to recovery but the accurate identification of recoverable and cost avoidable cases is problematic because many insurance plans use static lists of diagnosis codes that are based on the subjective assessment of exactly what constitutes a good code. Moreover, these lists are often not updated enough to ascertain the specific codes or combinations thereof that are pertinent to the best cases.

There is a better way. Just as credit card companies have been able to leverage Big Data to highlight suspicious activity and block card usage almost immediately, so too these techniques are making it possible to identify health insurance fraud in real time. It is now possible to use techniques such as predictive analytics and machine learning to reduce health insurance fraud, waste, and error by:

  • Identifying and maximizing fraud prevention and mistake prevention opportunities
  • Refining the fraud recovery process
  • Leveraging data analytics and on-demand metrics to improve program management
  • Implementing scoring and predictive analytics techniques that are superior to current flawed methods
  • Teaching your staff how to analyze data to better identify fraud and error

The ultimate goal is to create a "real-time" fraud detection model that allows for the identification and mitigation of fraud at the point of claims submission which is where consumer financial services companies have been moving towards. Experfy’s data scientists and consultants can help you to move towards this goal which is important for your company, your consumers and society.

Our Solution

Experfy consultants and data scientists have deep expertise in the design and implementation of health insurance fraud detection models that leverage the most innovative Big Data techniques such as predictive modeling and machine learning. The basis of our approach is to analyze your current fraud detection practices/technologies and to identify the bottlenecks that need to be addressed to reduce both fraud and error. Our mission is to minimize your fraud rate and increase your profits by lowering the costs of health insurance fraud for your organization. Who are your riskiest customers? What kinds of claims should be immediately flagged as suspicious? What data can likely be considered a mistake? How do we quickly alert decisionmakers to potential fraud, error and waste? What preventative technologies and business practices need to be in place?

Once these fundamental questions are addressed, our data scientists can help you to harness sophisticated Big Data techniques such as machine learning and predictive analytics to calculate the percentage likelihood that a claim is invalid based upon the use case and the behavior of specific customers. We can also leverage access to large data sets to help you to determine the amount that should be paid for a claim so as save money by reducing overpayments. In addition, given the large amount of data that health insurance firms must manage, we can help you to understand which modes of data storage and data organization are best for getting the most out of your existing data resources. Most importantly, we have the ability to use advanced machine learning to create customized health insurance fraud detection platforms that are HIPAA compliant; user-friendly; normalize data and enhance data quality; and detect fraud in real time.

Experfy’s real time fraud detection model is unique in that it is built upon an advanced machine learning platform.  It leverages a REST API that data scientists and developers can use to integrate behavioral data as it relates to users, items/use cases, and actions. This makes it ideal for situations that require highly personalized and customized outputs such as is the case in health insurance fraud. We have developed highly effective algorithms that can run our machine learning platform (or can be operationalized within your stack) and that are tailor-made for identifying health insurance fraud and medical errors. Experfy data scientists can also implement advanced scoring and predictive analytics to help you understand those specific diagnosis, revenue and procedure codes, populations, and practices that are most likely to result in recoveries. We are constantly analyzing these codes and other factors in the health care milieu that are essential to both recovery and loss prevention for health insurance plans. For example, our data scientists are adept at overcoming the bottlenecks posed by static lists of diagnosis codes through the use of table driven analysis whereby health insurance plans can mine data from those codes that are indicative of medical malpractice and flag them for reporting and tracking over time. The ability to maintain these histories and analyze the data within them is an essential part of our advantage.

The struggle against fraud is an ongoing one as new threats constantly emerge. To address that reality, we can help to train your personnel in the best managerial practices for making fraud detection and error reduction part of your business DNA and strategic approach. Health care has never been more important than ever so it is now targeted by hackers even more so than the data in retailers’ databases. Experfy can help you to adapt to this fundamental reality and thrive in it.

Tell Us About Your Specific Project

To assist our consultants and data scientists in developing a customized solution for you, please provide the following information:


  • What information is typically required in your patient claims (patient demographics, provider, place of service, type of service, diagnoses, procedures, etc.)?
  • What information is typically required in your provider claims (provider address, specialty, payment arrangement, etc.)?


  • What are your current fraud rates?
  • How much money do you lose to fraudulent claims?


  • What practices do you use to detect fraud and error?
  • What technology stack, if any, do you currently use as part of a fraud detection model?


  • What is the most important problem that you think we should address?

Cutting-Edge Fraud & Risk Analytics Expertise

Experfy provides the world's most prestigious talent on-demand

Works at Ernst & Young
Senior Enterprise Intelligence - Advanced Analytics
Worked at ING
Quantitative Risk Analyst
Worked at Enova Financial
Advanced Analytics Manager

Request a Free Consultation

Tell us about your business problem and we help you define it further.

The Harvard Innovation Lab

Made in Boston @

The Harvard Innovation Lab