Anti-Fraud with Hybrid Analytics

Fluent Analytics Fraud Waste and Abuse

Fraudulent claims in the U.S. carry a heavy price tag. Billions of dollars of fraudulent claims pass through the U.S Healthcare system every year. Fluent Analytics uses BI Tools and Anti-Fraud best practices to assist in uncovering of fraudulent claims, and provides robust solutions to detect where fraud is coming from.

Of the range of solutions that have been utilized by the larger Anti Fraud Industry, healthcare organizations are migrating from the 'pay and chase' method of detecting and recovering fraud, waste and abuse (FWA), to a more practical and refined model, which prioritize and enhance prevention. The strong belief is that effective use of more advanced data-sets and analytic technologies at the beginning of the process of reimbursement can supply the necessary security that the U.S. healthcare industry now requires to lessen FWA.

Data Mining

Data Mining has emerged as one of the most promising techniques to combat FWA.   Medicare along with Medicaid fraud is responsible for losses in the billions of dollars every year. Historically, Medicaid fraud had to depend on the Medicaid agency as well as other third party entities to analyze and classify possible cases of fraud. This type of reliance has severely hindered attempts to aggressively and efficiently chase down fraud. Anti-fraud organizations are crippled by procedures demanding manual intervention, and overburdened with a huge number of leads than they are capable of managing effectively. Staff limitations make the simple process of governing the leads and following up on them up quite staggering. Considerable amounts of time are wasted investigating records to determine the leads that are worth going after. Investigators are working under stressful conditions, attempting to keep up with the continuous inflow of new fraud cases and subsequent leads. These strained conditions can result in investigators overlooking less apparent connections in the data. Moreover, they may not get complete access to the entire data-set that might assist them with more relevant decision making. It has been almost impossible for investigators to maintain focus on the most extraordinary, high profile cases – until now.

Data Mining for Fraudulent Medicare Claims Example

  • Data mining for an “impossible day”, such as every weekend billed in a month.
  • Psychiatrist billed for 60 hours a day, or more than 50 patients per day.
  • Proactive analysis indicated individual psychiatric consultations billed at a high level only, excessively.
  • As a follow up to Data Mining results, clients found Patients interviewed never went to psychiatrist.

Our Approach

Fluent Analytics offers improvements in every aspect of FWA management through the application of automated procedures, and the use of well established data mining technologies coupled with hybrid analytic detection methods.

  • Utilizing hybrid analytics, the enhanced data mining can more effectively pinpoint fraud. Risk scores at the entity level can be developed using a scoring engine with the application of hybrid analytics; such as predictive models, business rules, anomaly detection, as well as social network analytics. The analytics can be adjusted according to the types of cases you wish to investigate, and further data can be incorporated that the Medicaid agency may not be utilizing or have available.
  • Prioritize and emphasize cases proficiently with the help of an intuitive analytical interface. Entity relationships are successfully identified by the system, pinpointing potential collusion or fraud rings. Next, it connects and gives priority to cases based on the entire network value and the associated combined risk.
  • Processes used to extract, clean and integrate data originating at different external and internal databases is automated. Information from the Medicaid claims system, all-payer claims database or clearinghouse, information of provider enrolment, evidence of criminal history, etc. can be coordinated and enriched automatically to add risk measures. Once the entity resolution is resolved we prioritize cases from a single user interface.