System and process for a fusion classification for insurance underwriting suitable for use by an automated system

• US 8,214,314 B2
• Filed: 06/02/2008
• Issued: 07/03/2012
• Est. Priority Date: 04/30/2003
• Status: Active Grant

• Current Assignee: Pacific Life Insurance Company (Pacific Mutual Holding Company)
• Original Assignee: Genworth Financial Incorporated
• Inventors: Bonissone, Piero Patrone, Aggour, Kareem Sherif, Subbu, Rajesh Venkat, Yan, Weizhong, Iyer, Naresh Sundaram, Chakraborty, Anindya
• Primary Examiner(s): Holmes, Michael B
• Application Number:
  US12/131,545
• Publication Number:
  US 20090048876A1
• Time in Patent Office: 1,492 Days
• Field of Search: 706 45- 47
• US Class Current: 706/46
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/08 Insurance

Abstract

A method and system for fusing a collection of classifiers used for an automated insurance underwriting system and/or its quality assurance is described. Specifically, the outputs of a collection of classifiers are fused. The fusion of the data will typically result in some amount of consensus and some amount of conflict among the classifiers. The consensus will be measured and used to estimate a degree of confidence in the fused decisions. Based on the decision and degree of confidence of the fusion and the decision and degree of confidence of the production decision engine, a comparison module may then be used to identify cases for audit, cases for augmenting the training/test sets for re-tuning production decision engine, cases for review, or may simply trigger a record of its occurrence for tracking purposes. The fusion can compensate for the potential correlation among the classifiers. The reliability of each classifier can be represented by a static or dynamic discounting factor, which will reflect the expected accuracy of the classifier. A static discounting factor is used to represent a prior expectation about the classifier'"'"'s reliability, e.g., it might be based on the average past accuracy of the model, while a dynamic discounting is used to represent a conditional assessment of the classifier'"'"'s reliability, e.g., whenever a classifier bases its output on an insufficient number of points it is not reliable.

Claims

1. A computer implemented system for underwriting an insurance application based on a plurality of previous insurance application underwriting decisions, the system comprising:

a plurality of decision engines where each decision engine generates an output result;

a fusion engine, where the fusion engine;

compares the output results of the plurality of decision engines; and

fuses the plurality of output results into a single fused decision;

a measurement engine for measuring at least a consensus of the fused decision;

an estimation engine to estimate the degree of confidence in the fused engine based on the measured consensus;

a production decision engine, where the production decision engine generates a production output result; and

a comparison engine, where the comparison engine compares the production output result with the fused decision; and

where each of the decision engine output results comprises a classification designation, and the fusion of the output results is based at least in part on a correlation between the classification designations of each of the output results.

2. The system according to claim 1, where:

the production output result comprises a classification designation for the insurance application and a confidence measure; and

the fused decision comprises a classification designation for the insurance application and a confidence measure.

3. The system according to claim 2, where the comparison of the results of the production decision engine with the fused decision includes a comparison of:

the production output result classification designation and the fused decision classification designation; and

the production output result confidence measure and the fused decision confidence measure; and

where the comparison engine generates a compared result comprising a classification designation and a confidence measure.

4. The system according to claim 1, where the fusion of the plurality of output results is an associative function.

5. The system according to claim 1, further comprising at least one processor engine for:

processing each of the plurality of decision engine output results; and

generating a processed output result for each of the plurality of decision engine output results, where the plurality of processed output results are compared and fused by the fusion engine.

6. The system according to claim 1, where each of the plurality of decision engine output results comprise:

at least one classification designation, where the classification designation comprises an assigned weight to at least one classification category, where the weight indicates a degree of preference for the classification category by the decision engine; and

a discounting factor, where the discounting factor includes an assigned weight to a discounting factor, where the weight indicates a degree of confidence in the at least one classification designation.

7. The system according to claim 6, where the discounting factor is one of a dynamic discounting factor and a static discounting factor.

8. The system according to claim 1, where the comparison engine further identifies the application for one of:

an audit;

augmentation of a training set;

review by an underwriter; and

recording of the applications'"'"' occurrence for tracking purposes.

9. A computer implemented process for underwriting an insurance application based on a plurality of previous insurance application underwriting decisions, the process comprising:

generating a plurality of decision output results for the application, where each of the decision output results is generated by a separate decision engine;

comparing the plurality of decision output results;

fusing the plurality of decision output results based at least in part on the comparison;

measuring the fused decision for at least a consensus;

estimating a degree of confidence in the fused decision based on the measured consensus;

generating a production output result;

comparing the production output result with the fused decision; and

where each of the decision output results comprises a classification designation, and the fusion of the decision output results is based at least in part on a correlation between the classification designations of each of the output results.

10. The process according to claim 9, where:

the production output result comprises a classification designation for the insurance application and a confidence measure; and

the fused decision comprises a classification designation for the insurance application and a confidence measure.

11. The process according to claim 10, where the comparison of the production output result and the fused decision includes a comparison of:

the production output result classification designation and the fused decision classification designation; and

the production output result confidence measure and the fused decision confidence measure; and

further comprising the step of generating a compared result comprising a classification designation and a confidence measure.

12. The process according to claim 9, where the fusion of the plurality of output results is an associative function.

13. The process according to claim 9, further comprising the steps of:

processing each of the decision output results; and

generating a processed output result for each of the plurality of decision output results, where the plurality of processed output results are compared and fused by the fusion engine.

14. The process according to claim 9, where each of the plurality of decision output results comprise:

at least one classification designation, where the classification designation comprises an assigned weight to at least one classification category, where the weight indicates a degree of preference for the classification category by the decision engine; and

a discounting factor, where the discounting factor includes an assigned weight to a discounting factor, where the weight indicates a degree of confidence in the at least one classification designation.

15. The process according to claim 14, where the discounting factor is one of a dynamic discounting factor and a static discounting factor.

16. The process according to claim 9, where the step of comparing further comprises identifying the application for one of:

an audit;

augmentation of a training set;

review by an underwriter; and

recording of the applications'"'"' occurrence for tracking purposes.

17. A computer implemented system that generates a classification designation based on a plurality of output results, the system comprising:

a plurality of decision engines where each decision engine generates an output result;

a fusion engine, where the fusion engine;

compares the output results of the plurality of decision engines; and

fuses the plurality of output results into a single fused decision;

a production decision engine, where the production decision engine generates a production output result;

a comparison engine, where the comparison engine compares the production output result with the single fused decision to generate a comparison result; and

an output portion that outputs the comparison result.

18. The system of claim 17, the system including:

a measurement engine that measures a consensus of the fused decision; and

an estimation engine that estimates the degree of confidence in the fused decision based on the measured consensus.

19. The system of claim 18, where each of the decision engine output results comprises a classification designation, and the fusion of the output results is based at least in part on a correlation between the classification designations of each of the output results.

20. The system of claim 19, wherein the classification designation is a classification designation for an insurance application.