Fraud detection in interactive voice response systems

An apparatus that determines a risk score of a call received by an Interactive Voice Response (IVR) system, the apparatus comprising:

• at least one processor;
• and
• a non-transitory computer readable medium coupled to the at least one processor having instructions stored thereon that, when executed by the at least one processor, causes the at least one processor to:
• create a feature vector based on interaction with the IVR system during the call;
• and
• use a machine learning model to determine the risk score of the call based on the created feature vector,
• wherein the created feature vector is based on a behavior feature of the call, and wherein the behavior feature of the call is a sequence of actions taken during interaction with an IVR system,
• wherein the behavior feature of the call is at least one of an action taken during interaction with an IVR system, an amount of time elapsed between actions taken during interaction with an IVR system, providing incorrect information to an IVR system, a number of times a specific activity in an IVR was performed, a number of times an IVR was called during a measure of time, a volume or a duration of at least one DTMF tone during interaction with an IVR system, an amount of time elapsed between DTMF tones, a use of voice during interaction with an IVR system, an amount of time elapsed between a beginning of an IVR prompt and a user's spoken response to the IVR prompt, or an amount of time elapsed between an IVR prompt and an action taken during interaction with an IVR system.