Most fraud-fighting tools fall into one of three categories:Rules-based systems, also known as expert systems, "score" the probability that a claim has some element of fraud. Data elements from a claim-names, addresses, descriptions of damages or injuries-are run against a pre-defined set of fraud indicators. The downside of these systems is they generate many false-positive results.
Predictive modeling also scores claims on the probability of fraud, but instead of rules, it uses a carrier's own historical claims and fraud investigation data. Complex algorithms can be developed for each line of business. The models "predict" the probability of fraud based on past claims that were investigated for fraud.
Data mining and visualization tools are used by investigators to link what to the eye appears to be unrelated data and transforms it into useful information. Visualization software is being used to identify patterns, trends and relationships in data from multiple sources, such as mapping a company's historical claims with industry claims data.
