Our client, a private health insurance company, has about two million policy holders. The company used to rely on superficial processes to detect cases of fraud in billing. For example, the clerks checked whether the medication listed on the bill submitted is still being produced. This inadequate verification system made things easy for dishonest customers. The costs of such fraud cases are borne by everyone.

The adesso experts collaborated with the customer to analyse how AI applications can improve these processes. The aim was to automate parts of the inspection process, increase the detection rate and reduce the damage. In the first step, the experts developed a model to detect aberrant customer behaviour. The idea was to be able to identify fraud cases in a targeted manner. To this end, the project team performed an ad-hoc analysis to ascertain the feasibility. Following this, they developed a prototype model.

The AI application automatically filters out conspicuous events from the large number of billing statements. These are then submitted to the clerks for review. The experts perform a detailed examination, research the connections and, if necessary, talk with the parties involved. They focus on cases with a high probability of irregularities.