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

• US 20030158751A1
• Filed: 11/15/2002
• Published: 08/21/2003
• Est. Priority Date: 07/28/1999
• Status: Active Grant

• Current Assignee: Fair Isaac Corporation
• Original Assignee: Fair Isaac Corporation
• Inventors: de Traversay, Jean, Suresh, Nallan C., Pathria, Anu K., Gollamudi, Hyma, Tyler, Michael K.
• Granted Patent: US 8,666,757 B2
• Time in Patent Office: Days
• US Class Current: 705/2
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0283 Price estimation or determination
• CPC Class Codes: G06Q 30/0603 Catalogue ordering
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03

Abstract

Fraud and abuse detection in an entity'"'"'s payment coding practices includes the ability to search for fraud at all levels of the hierarchical coded payment system within the context of an unsupervised model. The model uses variables derived and profiles created at any level or at all levels of the hierarchical coded payment system to create a comprehensive description of the payment coding activities submitted by the entity. That description is compared with other peer entities to determine unusual and potentially inappropriate activity. The profiles created may themselves be utilized for purposes other than the detection of fraud and abuse.

Claims

1. A computer-implemented method for evaluating the behavior of a facility from the facility'"'"'s activity in a hierarchical coded payment system, the method comprising:

calculating summary variables from data for a particular activity 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;

determining normalized variables based on comparing the summary variables with peer data for the particular metric; and

deriving a behavior indicator from the normalized variables, the indicator indicating a measure of aberrance of the facility'"'"'s behavior with respect to the peer data.

2. The computer-implemented method according to claim 1, wherein the hierarchical coded payment system is selected from a group of Prospective Payment Systems (PPS) comprising Medicare Ambulatory Surgical Center PPS, Medicare Inpatient Hospital PPS, Medicare Skilled Nursing Facility PPS, Medicare Home Health PPS, Medicare Outpatient Hospital PPS, Medicare Inpatient Rehabilitation Facility PPS, Medicare Part C risk adjustment, Medicare Swing Bed Facility PPS, Medicare Long-Term Care PPS, any future Medicare Part B procedure PPS, Medicaid PPS, private insurer'"'"'s PPS (including private version of the CMS PPS), and national-payer healthcare PPS.

3. The computer-implemented method according to claim 1, wherein the data is obtained in batches from transactional level data associated with the facility.

4. The computer-implemented method according to claim 1, wherein the data is obtained from updates made to transactional level data associated with the facility.

5. The computer-implemented method according to claim 1, wherein the hierarchical coded payment system includes a plurality of classification levels defining the payment determined, the plurality of classification levels comprising:

a primary level including a set of driving elements used to encode the service provider activity at a transactional level;

an intermediary level including a set of groups, each group mapping one or more driving elements to a particular payment rate; and

an aggregate level including a set of categories, each category being mapped to one or more of the groups according to predetermined industry classification schemes.

6. The computer-implemented method according to claim 5, wherein the hierarchical coded payment system comprises a Medicare Inpatient Hospital Prospective Payment System, the driving elements comprise Diagnosis Codes, the groups comprise Diagnosis Related Groups, and the categories comprise Major Disease Categories.

7. The computer-implemented method according to claim 5, wherein the driving elements comprise Principal Diagnosis codes.

8. The computer-implemented method according to claim 5, wherein the hierarchical coded payment system comprises a Medicare Skilled Nursing Facility Prospective Payment System, the driving elements comprise a Minimum Data Set, the groups comprise Resource Utilization Groups, and the categories comprise Major Resource Categories.

9. The computer-implemented method according to claim 5, wherein summary variables comprise one of the data extracted across the primary level, the data extracted within the driving elements, the data extracted across the intermediary level, the data extracted within the groups, the data extracted across the aggregate level, and the data extracted within the categories.

10. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics across the primary level into a profile; and

deriving the summary variables from the profile.

11. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics within the driving elements into a profile; and

deriving the summary variables from the profile.

12. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics across the intermediary level into a profile; and

deriving the summary variables from the profile.

13. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics within the groups into a profile; and

deriving the summary variables from the profile.

14. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics across the aggregate level into a profile; and

deriving the summary variables from the profile.

15. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics within the categories into a profile; and

deriving the summary variables from the profile.

16. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics across the facility into a profile; and

deriving the summary variables from the profile.

17. The computer-implemented method according to claim 5, wherein calculating summary variables from the data comprises:

capturing behavioral characteristics within the facility into a profile; and

deriving the summary variables from the profile.

18. The computer-implemented method according to claim 1, wherein comparing the summary variables with industry-wide peer data for the particular metric comprises:

determining a first distribution based on the summary variables;

determining a second distribution based on the industry-wide peer data;

detecting aberrations between the first distribution and the second distribution; and

integrating the aberrations detected into the normalized variables.

19. The computer-implemented method according to claim 1, wherein determining normalized variables comprises:

merging the summary variables with the industry-wide peer data; and

rolling-up the summary variables with the industry-wide peer data.

20. The computer-implemented method according to claim 19, wherein rolling-up the summary variables comprises:

applying a distributional function to the metric across all of the summary variables; and

responsive to distributional function applied, determining a scalar quantity representing the normalized variables

21. The computer-implemented method according to claim 1, wherein the metric is selected from one of a group of metrics comprising an indicator of total costs claimed by the facility, an indicator of the facility'"'"'s average patient length of stay, and an indicator of a number of claims made by the facility.

22. The computer-implemented method according to claim 1, wherein deriving an indicator from the normalized variables comprises:

determining a score value for the normalized variables;

producingreasons derived from one or more top significant variables supporting the score; and

associating a threshold value with the score value, the indicator representing the potentially fraudulent service provider activity when a score value exceeds the threshold value.

23. The computer-implemented method according to claim 1, wherein potentially fraudulent service provider activity comprises the facility upcoding the payment in return for the services.

24. The computer-implemented method according to claim 1, wherein the indicator comprises a discrepancy between the summary variables compared with the peer data for the particular metric.

25. The computer-implemented method according to claim 1, wherein potentially fraudulent service provider activity comprises the facility causing inappropriate selection of the metric to obtain an increased amount of the payment.

26. The computer-implemented method according to claim 1, wherein potentially fraudulent service provider activity comprises the facility furnishing the services at a reduced level which is not commensurate with the payment.

27. The computer-implemented method according to claim 1, wherein the potentially fraudulent service provider activity comprises the facility furnishing the services fictitiously.

28. The computer-implemented method according to claim 1, wherein the facility is selected from a group of entities comprising healthcare related facilities, healthcare providers, patients, beneficiaries, healthcare claims processors, and skilled nursing facilities.

29. 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.

30. The computer-implemented method according to claim 29, wherein the hierarchical coded payment system comprises a Medicare Inpatient Hospital Prospective Payment System, the driving elements comprise Diagnosis Codes, the groups comprise Diagnosis Related Groups, and the categories comprise Major Disease Categories.

31. The computer-implemented method according to claim 29, wherein the driving elements comprise Principal Diagnosis codes.

32. The computer-implemented method according to claim 29, wherein the hierarchical coded payment system is selected from a group of Prospective Payment Systems (PPS) comprising Medicare Ambulatory Surgical Center PPS, Medicare Inpatient Hospital PPS, Medicare Skilled Nursing Facility PPS, Medicare Home Health PPS, Medicare Outpatient Hospital PPS, Medicare Inpatient Rehabilitation Facility PPS, Medicare Part C risk adjustment, Medicare Swing Bed Facility PPS, Medicare Long-Term Care PPS, any future Medicare Part B procedure PPS, Medicaid PPS, private insurer'"'"'s PPS (including private version of the CMS PPS), and national-payer healthcare PPS.

33. The computer-implemented method according to claim 29, wherein the hierarchical coded payment system comprises a Medicare Skilled Nursing Facility Prospective Payment System, the driving elements comprise a Minimum Data Set, the groups comprise Resource Utilization Groups, and the categories comprise Major Resource Categories.

34. The computer-implemented method according to claim 29, wherein the summary variables comprise one of the data extracted across the primary level, the data extracted within the driving elements, the data extracted across the intermediary level, the data extracted within the groups, the data extracted across the aggregate level, and the data extracted within the categories.

35. The computer-implemented method according to claim 29, wherein the normalized variables include a deviation measure based on the summary variables compared with the industry-wide peer data.

36. 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.

37. The computer-implemented method of claim 36, wherein the payment is determined according to a payment function associated with the PPS.

38. The computer-implemented method of claim 36, wherein PPS comprises a plurality of classification levels defining the payment, 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.

39. The computer-implemented method according to claim 38, wherein the summary variables comprise one of summary variables calculated across the driving element level, calculated within the driving elements, calculated across the group level, calculated within the groups, calculated across the category level, and calculated within the categories.

40. 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.

41. 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.

42. The neural network according to claim 41, wherein the PPS is defined according to a structure including

a set of driving elements, each driving element being used to encode service provider activities performed at a transactional level;

a set of groups, each group including one or more driving elements as a collection used to characterize a set of the activities, the groups enabled to map the service provider activities to a payment function; and

a set of categories, each category being mapped to one or more of the groups according to a predetermined set of classification schemes.

43. 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.

44. The computer program product according to claim 43, wherein the hierarchical coded payment system comprises a Prospective Payment System (PPS).

45. The computer program product according to claim 44, wherein the PPS is selected from one of a group of PPS'"'"' comprising Medicare Ambulatory Surgical Center PPS, Medicare Inpatient Hospital PPS, Medicare Skilled Nursing Facility PPS, Medicare Home Health PPS, Medicare Outpatient Hospital PPS, Medicare Inpatient Rehabilitation Facility PPS, Medicare Part C risk adjustment, Medicare Swing Bed Facility PPS, Medicare Long-Term Care PPS, Medicare Part B procedure PPS, Medicaid PPS, private insurer'"'"'s PPS, and national-payer healthcare PPS.

46. 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.

47. 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.