Health care fraud affects more than the bottom line. In the U.S. alone, health care fraud is estimated to cost somewhere between $80 billion and $170 billion per year, according to the National Health Care Anti-Fraud Association. Aside from the fact that fraud is a criminal activity and that taxpayers and government agencies are affected, the human factors alone are compelling enough for health care payers to take this issue seriously and develop a proactive approach to fraud detection. In some cases of fraud, people lose their life savings. Within health care related fraud, there is the potential to affect a patient’s health or future treatment options.


Billing for services never rendered and charging for more expensive procedures are just two ways that fraudulent health care providers affect patients. The sad fact remains that health care providers have also been known to perform unnecessary medical services for the sole purpose of collecting insurance payments. In addition, fraudulent providers falsify medical treatment histories or diagnoses of medical conditions and use up patients’ health care benefits, putting people’s lives at risk. The possibility to drain a patient’s private insurance benefits means that when they might really be needed, a patient may not have access to the appropriate insurance amounts required for adequate treatment. If a patient’s medical insurance is depleted, that may affect future treatments and in serious cases lead to premature death.


For health care payers, the long-term benefits of implementing a fraud detection solution offset high the initial implementation costs. As Anu Pathria, vice president of health care analytics at Fair Isaac succinctly states, “a predictive analytics system can holistically look at all of the different aspects of activity of the provider and bring it all together into a single synthesized fraud risk assessment.”


The Use of Predictive Analytics


Predictive analytics uses pattern identification to identify suspicious activity. This allows fraud to be detected before it happens, limiting the number of fraudulent claims that are paid and identifying potential future fraud before it occurs. The ability to detect subtle patterns within large datasets highlights the nature of predictive analytics. Each claim is tracked along with its details to identify subtleties that individuals may overlook or that would not be obvious. This may include overactive claims submissions, similar procedures being reported by a particular health care provider or patient claims being submitted with unrelated procedures listed.


Predictive analytics enables provider information to be collected over a long period of time to identify discrepancies, changes in submissions and general comparisons by looking at all of the information available. By using a data-driven approach and different predictive modeling techniques, overall claim data can be gathered and analyzed to identify problematic activity. For instance, deviation analysis can be applied to identify the providers who submit a higher amount of claims or who perform a higher percentage of tests and apply the deviation amounts by comparing the numbers to regular claim submission amounts. This enables health care payers to determine overall averages and to identify other health care providers who fall outside the standard submission patterns, flagging those providers. In general, with the amount of information collected – patient medical and treatment history, provider submission patterns, etc. – for a health care payer finding fraudulent activities alone is almost impossible.


Due to the complexities of detection, various predictive modeling methods can be applied to identify the probability of fraudulent activities. The combination of claim data and patient history allows health care payers to identify claim submissions that do not match the medical records of the patient. Although this might not always signify a fraudulent act, the claim still warrants extra investigation. This, coupled with the identification of unnatural similarities within the submission of various claims may flag unusual activity. In many cases, a health care provider will submit similar claims creating subtle submission patterns that can be detected over time.


If the estimates of health care fraud are correct, and three to 10 percent of claims submitted are fraudulent, then detecting even one percent of the 10 percent can save health care payers $150 million annually. By taking a proactive approach to detecting fraud, providers may be less likely to risk submitting fraudulent claims if there are known, formalized detection procedures for every claim submitted. In some cases, patients have also been given control over their own insurance to shift the responsibility and to help minimize dishonest claims from being submitted.


No One-Stop Shop Solution


Unfortunately, health care and the processes associated with patient care differ by region (state, country, etc.). Fraud detection solutions should take into account local processes, insurance structures and government legislation. Whether private insurance agencies or governments are responsible for the bulk of health care claims payment, the identification of fraudulent claims before they are paid should be a primary goal on the road to early detection.


For more insights into predictive analytics and fraud detection for health care payers, please listen to the following podcast with Anu Pathria at Fair Isaac Corporation.


Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access