Fraud and abuse detection and entity profiling in hierarchical coded payment systems

A computer-implemented method for determining potentially fraudulent service provider activity in a hierarchical coded payment system, the method comprising:

• obtaining data representing services provided by at least one facility in return for payment, the payment determined using the hierarchical coded payment system, having a plurality of classification levels defining the payment determined, the plurality of classification levels comprising,
• a driving element level including a set of driving elements used to encode the service provider activity at a transactional level,
• a group level including a set of groups, each group mapping one or more driving elements to a particular payment rate, and
• a category level including a set of categories, each category being mapped to one or more of the groups according to predetermined industry classification schemes;
• calculating summary variables from the data for a particular metric associated with the hierarchical coded payment system;
• determining normalized variables based on comparing the summary variables with industry-wide peer data for the particular metric;
• and
• deriving an indicator from the normalized variables, the indicator representing the potentially fraudulent service provider activity.

A computer-implemented method for generating fraud indication within a Prospective Payment System (PPS), the method comprising:

• generating profiles of service provider activities rendered for payment by a facility, the profiles being dynamically derived from transactional level data associated with service provider activities;
• calculating summary variables from the profiles input into a predictive model for a particular metric associated with the PPS;
• determining a deviation measure based on comparing the summary variables with industry-wide peer data for the particular metric;
• and
• deriving an indicator from the deviation measure, the indicator representing the fraud indication based on aberrations associated with the deviation measure.

A computer-implemented method for determining potentially fraudulent service provider activity in a hierarchical coded payment system, the method comprising:

• obtaining data representing services provided by a facility in return for payment, the payment determined using the hierarchical coded payment system having
• a driving element level including a set of driving elements used to encode the service provider activity at a transactional level, and
• a group level including a set of groups, each group mapping one or more driving elements to a particular payment rate;
• identifying a driving element set comprising a plurality of groups to which a plurality of driving elements map thereto;
• identifying all combinations of pairs of groups within the driving element set;
• for each pair, calculating summary variables from the data for a particular metric associated with the hierarchical coded payment system;
• within the pair, determining normalized variables based on comparing the summary variables for both groups in the pair with industry-wide peer data;
• and
• deriving indicators of the potentially fraudulent service provider activity from the normalized variables representing a group in the pair that is associated with a higher payment rate.

In a computer-controlled Prospective Payment System (PPS) including a computer readable memory and a neural network stored in the computer readable memory, the neural network detecting potentially fraudulent service provider activity in the PPS, comprising:

• a first calculator capable of producing profiles from claim data and summary variables encoded from transaction data associated with the PPS;
• coupled to the first calculator, a second calculator capable of producing industry-wide statistical peer data;
• coupled to the second calculator, a generator enabled to provide a deviation measure based on comparing the profiles with industry-wide peer data;
• and
• coupled to the generator, an indicator capable of detecting the potentially fraudulent service provider activity based on aberrations associated with the deviation measure.

A computer program product for determining potentially fraudulent service provider activity in a hierarchical coded payment system, the program product stored on a computer readable medium and adapted to perform the operations of:

• allowing calculation of summary variables from data for a particular metric associated with the hierarchical coded payment system, the data representing services provided by at least one facility in return for payment determined using the hierarchical coded payment system;
• allowing normalized variables to be determined based on comparing the summary variables with industry-wide peer data for the particular metric;
• and
• enabling derivation of an indicator from the normalized variables, the indicator representing the potentially fraudulent service provider activity.

A computer program product for determining potentially fraudulent service provider activity in a hierarchical coded payment system, the program product stored on a computer readable medium and adapted to perform the operations of:

• allowing data to be obtained representing services provided by at least one facility in return for payment, the payment determined using the hierarchical coded payment system, having a plurality of classification levels defining the payment determined, the plurality of classification levels comprising,
• a driving element level including a set of driving elements used to encode the service provider activity at a transactional level,
• a group level including a set of groups, each group mapping one or more driving elements to a particular payment rate, and
• a category level including a set of categories, each category being mapped to one or more of the groups according to predetermined industry classification schemes;
• enabling calculation of summary variables from the data for a particular metric associated with the hierarchical coded payment system;
• enabling determination of normalized variables based on comparing the summary variables with industry-wide peer data for the particular metric;
• and
• allowing derivation of an indicator from the normalized variables, the indicator representing the potentially fraudulent service provider activity.

A computer-implemented method for evaluating an entity, wherein the entity has activities or attributes which are classified in a hierarchical classification scheme from transactions associated with the entity, each classification associated with a quantitative value, the method comprising:

• generating a profile of the entity's activities or attributes based on transaction level data from the entity's transactions, and derived from the quantitative values associated with the classifications of the activities or attributes in the entity's transactions;
• calculating summary variables from the profile;
• normalizing the summary variables with respect to variables derived from the activities or attributes of peers of the entity;
• scoring the normalized profile of the entity using an unsupervised predictive model of a selected metric associated with the hierarchical classification scheme, to produce a deviation measure;
• and
• deriving an indicator from the deviation measure, the indicator representing the evaluation of the entity based on the deviation measure.