Method of personalizing, individualizing, and automating the management of healthcare fraud-waste-abuse to unique individual healthcare providers

Patent metadata

Field	Value
Publication number	US-12437343-B2
Application number	US-202418437315-A
Country	US
Kind code	B2
Filing date	Feb 9, 2024
Priority date	Oct 15, 2014
Publication date	Oct 7, 2025
Grant date	Oct 7, 2025

Abstract

A method of preventing healthcare fraud-waste-abuse uses artificial intelligence machines to limit financial losses. Healthcare payment request claims are analyzed by predictive models and their behavioral details are compared to running profiles unique to each healthcare provider submitting the claims. A decision results that the instant healthcare payment request claim is or is not fraudulent-wasteful-abusive. If it is, a second analysis of a group behavioral in which the healthcare provider is clustered using unsupervised learning algorithms and compared to a running profile unique to each group of healthcare providers submitting the claims. An overriding decision results if the instant healthcare payment request claim is or is not fraudulent-wasteful-abusive according to group behavior.

First claim

The invention claimed is:

1. A computer-implemented method for healthcare fraud detection, comprising, via one or more processors:
receiving a healthcare payment request corresponding to a healthcare provider;
matching the healthcare payment request to a smart agent having a profile comprising a representation of historical data of the healthcare provider;
analyzing the healthcare payment request for fraud with a predictive model constructed according to one or more of: data mining logic, a neural network, case-based-reasoning, clustering, fuzzy logic, a genetic algorithm, a decision tree, and business rules;
comparing the contents of at least one data field of the healthcare payment request against the profile of the corresponding smart agent and, based at least in part on the comparison and on the analysis under the predictive model, generating a first output indicative of a determination of likely fraud;
generating a second smart agent having a second profile by supervised clustering of a group of healthcare providers based on at least one claim attribute;
matching the healthcare payment request to the second smart agent having the second profile;
comparing the contents of at least one data field of the healthcare payment request against the second profile of the corresponding second smart agent and, based at least in part on the comparison, generating a second output indicative of a determination of unlikely fraud;
analyzing the first output and the second output and generating a decision on the healthcare payment request based at least in part on the analysis, the analysis including determining that the first output meets a first criteria indicating fraud or abuse, determining that the second output meets a second criteria indicating acceptable behavior within the corresponding group, and determining that the second output overrides the first output,
wherein the decision comprises an approval decision for the healthcare payment request.

2. The computer-implemented method of claim 1 , wherein the second smart agent is generated in part by unsupervised clustering.

3. The computer-implemented method of claim 1 , further comprising updating, via the one or more processors, the second profile to reflect the second output.

4. The computer-implemented method of claim 1 , further comprising—
inputting the contents of at least one data field of the healthcare payment request to a treatment model to produce a predicted diagnostic code,
comparing the predicted diagnostic code to the healthcare payment request to produce a third output, the decision being based at least in part on the third output.

5. The computer-implemented method of claim 4 , wherein—
comparing the predicted diagnostic code to the healthcare payment request includes determining whether the predicted diagnostic code matches one or both of a prescription and procedure code corresponding to the healthcare payment request,
the third output is indicative of a determination that the predicted diagnostic code does not match one or both of the corresponding prescription and procedure codes.

6. The computer-implemented method of claim 4 , wherein—
comparing the predicted diagnostic code to the healthcare payment request includes determining whether the predicted diagnostic code matches one or both of an expected diagnostic test and diagnostic test results corresponding to the healthcare payment request,
the third output is indicative of a determination that the predicted diagnostic code matches one or both of the corresponding expected diagnostic test and diagnostic test results.

7. The computer-implemented method of claim 6 , wherein the third output contributes to the override of the first output.

8. At least one network server for healthcare fraud detection, the at least one network server comprising:
one or more processors;
non-transitory computer-readable storage media having computer-executable instructions stored thereon, wherein when executed by the one or more processors the computer-readable instructions cause the one or more processors to—
receive a healthcare payment request corresponding to a healthcare provider;
match the healthcare payment request to a smart agent having a profile comprising a representation of historical data of the healthcare provider;
analyze the healthcare payment request for fraud with a predictive model constructed according to one or more of: data mining logic, a neural network, case-based-reasoning, clustering, fuzzy logic, a genetic algorithm, a decision tree, and business rules;
compare the contents of at least one data field of the healthcare payment request against the profile of the corresponding smart agent and, based at least in part on the comparison and on the analysis under the predictive model, generate a first output indicative of a determination of likely fraud;
generate a second smart agent having a second profile by supervised clustering of a group of healthcare providers based on at least one claim attribute;
match the healthcare payment request to the second smart agent having the second profile;
compare the contents of at least one data field of the healthcare payment request against the second profile of the corresponding second smart agent and, based at least in part on the comparison, generate a second output indicative of a determination of unlikely fraud;

analyze the first output and the second output and generate a decision on the healthcare payment request based at least in part on the analysis, the analysis including determining that the first output meets a first criteria indicating fraud or abuse, determining that the second output meets a second criteria indicating acceptable behavior within the corresponding group, and determining that the second output overrides the first output,
wherein the decision comprises an approval decision for the healthcare payment request.

9. The at least one network server of claim 8 , wherein the second smart agent is generated in part by unsupervised clustering.

10. The at least one network server of claim 8 , wherein, when executed by the one or more processors, the computer-readable instructions further cause the one or more processors to update the second profile to reflect the second output.

11. The at least one network server of claim 8 , wherein, when executed by the one or more processors, the computer-readable instructions further cause the one or more processors to—
input the contents of at least one data field of the healthcare payment request to a treatment model to produce a predicted diagnostic code,
compare the predicted diagnostic code to the healthcare payment request to produce a third output, the decision being based at least in part on the third output.

12. The at least one network server of claim 11 , wherein—
comparing the predicted diagnostic code to the healthcare payment request includes determining whether the predicted diagnostic code matches one or both of a prescription and procedure code corresponding to the healthcare payment request,
the third output is indicative of a determination that the predicted diagnostic code does not match one or both of the corresponding prescription and procedure codes.

13. The at least one network server of claim 11 , wherein—
comparing the predicted diagnostic code to the healthcare payment request includes determining whether the predicted diagnostic code matches one or both of an expected diagnostic test and diagnostic test results corresponding to the healthcare payment request,
the third output is indicative of a determination that the predicted diagnostic code matches one or both of the correspondin

Assignees

• Brighterion Inc

Inventors

• Adjaoute Akli

Classifications

• G06Q10/10

  Office automation; Time management · CPC title

• Y02A90/10

  Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation · CPC title

• G06N20/00

  Machine learning · CPC title

• G16H40/20

  for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms · CPC title

• G06Q40/08 Primary

  Insurance · CPC title