Healthcare fraud is a serious issue in the U.S., one that results in annual financial losses of tens of billions of dollars. This dishonest practice not only increases the costs of premiums for patients but also puts their lives at risk.
However, keeping track of thousands of claims, providers, and patients is an incredibly daunting task for any individual investigator. Fortunately, there exist a number of intelligent counter fraud solutions that help data experts identify suspicious scenarios more easily. These solutions can be integrated into existing systems with little to no disruption, and allow health payers to take advantage of predictive analytics in the following ways:
Discover Unusual Emergent Behaviors
In most cases, subject matter experts outline the specific rules or scenarios that determine whether certain transactions must be investigated for possible abuse. However, the nature of this set-up means that it can only predict previously identified types of fraud scenarios. Predictive analytics gets around this by using cluster analysis, a model that assembles groups of similar claims, providers, and patients based on predetermined parameters, such as normal billing behaviors. This process can also conduct a link analysis, another process that pinpoints unusual connections between providers, suppliers, employees, and known criminals. These solutions can create a visual map of this resulting network of relationships, allowing experts to immediately spot new and unknown criminal patterns.
Predict Potential Fraud Scenarios
Uncovering unusual behavior and connections is only the first step of the process. Most counter-fraud solutions use predictive modeling to create a working statistical model based on historical data. These models are applied to incoming claims to determine the probability of criminal behavior. By creating robust groupings of claims, providers, and patients as well as creating and applying multiple models to them, these solutions can quickly pinpoint the tell-tale signs of known fraud scenarios. These allow experts to make accurate predictions on which claims may be false, both for pre-payment and post-payment transactions.
Explain Anomalous Behavior
An entity’s historical billing behavior often looks normal until a spike occurs, which is usually an indication of identity theft. However, there are times when sudden changes in entity behavior may not necessarily be a sign of suspicious activity. Counter-fraud solutions make use of spike analysis that visually present obvious and unusual actions. These are then compared with historical data, such as the time of year and location of the transaction, in order to determine whether said spike may be indicative of criminal behavior. Predictive analytics can also use said data to explain why some segmentation or behavior anomalies may exist without necessarily being false or abusive, leading to fewer false positive results.
Promotes Immediate Action
Determining which cases to investigate can be an incredibly taxing and resource-intensive task for experts to manually carry out. Luckily, counter-fraud solutions can be integrated into a payer’s analytical workflow. Using risk scoring—a predictive model that analyses available public and private data—these types of software can rank providers according to their risk level and create a prioritized list of claims to investigate. These solutions can also push all of this information to the appropriate teams or applications for immediate action. Because these solutions can directly pass on data to experts, this allows them to identify emerging fraud patterns at a glance and narrow their list of suspects down.
Learn from Existing Data
Because healthcare fraud strategies are always evolving, it is important to use software that can study incoming data and identify changing trends. Machine intelligence solutions use a model called trend analysis in order to spot patterns in unusual behavior, including incredibly subtle ones that may slip past an expert, and predict whether these will result in false claims. Predictive analytics software can then suggest if existing rankings and segments of entities should be updated to account for the identified emergent behavior.
Though many efforts have been made to catch false claims, it’s clear that the “pay and chase” model is a largely ineffective method of solving the problem. Instead of making a move only after the damage has already been done, it’s best to predict potential fraudulent activity and prevent it from occurring in the first place. By taking advantage of predictive analytics, health payers can not only cut costs, but they can also save thousands of patients from being exploited by hospitals and criminal groups.
