System for selection of data records containing structured and unstructured data

• US 10,445,697 B2
• Filed: 11/26/2012
• Issued: 10/15/2019
• Est. Priority Date: 11/26/2012
• Status: Active Grant

• Current Assignee: Hartford Fire Insurance Company (The Hartford Financial Services Group, Inc.)
• Original Assignee: Hartford Fire Insurance Company (The Hartford Financial Services Group, Inc.)
• Inventors: Li, Bixia, Drennan, III, Arthur Paul, Gooley, Brian J.
• Primary Examiner(s): Anderson, Mike
• Application Number:
  US13/684,783
• Publication Number:
  US 20140149144A1
• Time in Patent Office: 2,514 Days
• Field of Search: 705 36
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06Q 10/10 Office automation; Time management

Abstract

A computer system for processing data relating to injury claims to determine suitability for subrogation includes one or more data storage devices storing structured and unstructured data relating to insurance claims, a processor configured to access the one or more data storage devices to select claims for review, apply a text mining tool to extract data from the unstructured data associated with the selected claims, and apply a predictive model to the unstructured data and structured data associated with the selected claims to identify a likelihood of suitability for subrogation of each of the selected claims.

Claims

1. A computer system, comprising:

one or more data storage devices storing;

a plurality of data records, each data record comprising stored data, the stored data comprising;

structured data and unstructured data, the unstructured data comprising data indicative of communications between two or more persons;

a data storage device storing data defining a predictive model for determining a score indicative of suitability of data records for forwarding to an operation system, responsive to receipt of one of the data records, the predictive model configured to perform the determination based on review of the structured data and words and phrases extracted from the data indicative of the unstructured data of the data record, the predictive model comprising data values and coefficients associated with structured data, certain data values being negative and indicative of a reduced suitability for forwarding, and certain data values being positive and indicative of an increased suitability for forwarding;

an analysis hardware server, in communication with the data storage devices and comprising a processor configured to, on a recurring and dynamic basis, in each of a plurality of cycles;

determine a set of data records, from the plurality of data records, for review by the predictive model, the set comprising data records having no score determined in any prior cycle and selected data records assigned scores in one or more prior cycles, the selected data records assigned scores in one or more prior cycles comprising data records not forwarded to the operation system and having selected aging intervals;

extract words and phrases from the unstructured data associated with the determined set of data records using data mining to identify words and phrases for extraction, comprising using text parsing logic to identify sentences, and tagging identified words to be associated with categories and subcategories;

apply the predictive model to the structured data and the extracted words and phrases from the unstructured data to determine a current-cycle score associated with each of the data records in the set, and assign each of the data records of the set to one of a plurality of groups of equal numbers ordered in accordance with the determined current-cycle score;

for each of the data records of the set for which a score was determined in a prior cycle, determine whether the current-cycle score is greater than a most-recently determined prior cycle score for the same data record; and

generate a report including at least some of the data records having no score determined in any prior cycle and the data records assigned scores in one or more prior cycles determined to have a greater current-cycle score than the most-recently determined prior cycle score for the same data record;

determine a velocity of change in group to which each data record is assigned;

based at least in part on the determined velocity of change in group to which a data record is assigned, and based on whether a current-cycle score is at least a threshold value, select a plurality of the data records for forwarding to an operation system, wherein a positive velocity of change in group renders a data record more likely to be selected for forwarding, and a negative velocity of change in group renders a data record less likely to be selected for forwarding; and

forward, via bulk download using xml documents, to the operation system, the data records selected for forwarding.

2. The computer system of claim 1, wherein the data records comprise data indicative of workers compensation claims, and the data indicative of communications comprises at least notes based on conversations between claim handlers and one or more of injured individuals and employer representatives.

3. The computer system of claim 1, wherein the generating a report comprises grouping claims by score.

4. The computer system of claim 1, wherein the selected data records assigned scores in one or more prior cycles comprise data records not forwarded to the operation system and having selected aging intervals.

5. The computer system of claim 1, wherein the data records comprise data indicative of workers compensation claims and the structured data values comprise an on premises indicator having a positive or negative value and nature of injury indicator.

6. The computer system of claim 1, wherein the structured data values include a report lag indicator, having a positive coefficient for a short report lag value and a negative coefficient for a long report lag value.

7. The computer system of claim 1, wherein the predictive model comprises a plurality of terms capable of extraction from unstructured data and positive and negative coefficients associated with the terms.

8. The computer system of claim 7, wherein the terms comprise contractor and landlord, a positive coefficient being associated with those terms.

9. The computer system of claim 7, wherein the data records comprise workers compensation claims and one or more of the coefficients are employer description dependent.

10. The computer system of claim 1, further comprising a claim recovery operation hardware server in communication with the processor and configured to receive claim and likelihood of subrogation data.

11. A computer-implemented method, comprising:

storing in one or more data storage devices a plurality of data records, each of the data records comprising structured data and unstructured data, the unstructured data comprising data indicative of communications between two or more persons;

on a recurring and dynamic basis on a cycle accessing by a processor of an analysis hardware server from the one or more data storage devices data relating to the data records to determine a set of data records for review, the set of data records including data records having no score determined in any prior cycle and data records assigned scores in one or more prior cycles, the selected data records assigned scores in one or more prior cycles comprising data records having selected aging intervals;

processing by the processor using data mining the unstructured data to extract words and phrases from the unstructured data, comprising using text parsing logic to identify sentences, and tagging identified words to be associated with categories and subcategories;

applying, by the processor, to each of the data records of the set, a predictive model for determining a current-cycle score indicative of suitability of data records for forwarding to an operation system, the predictive model being configured to perform the determination responsive to receipt of one of the data records based on review of the structured data and words and phrases extracted from the unstructured data, to the structured data and the extracted words and phrases from the unstructured data of the data record, the predictive model comprising data values and coefficients associated with structured data certain data values being negative and indicative of a reduced suitability for forwarding, and certain data values being positive and indicative of an increased suitability for forwarding, and assigning each of the data records of the set to one of a plurality of groups of equal numbers ordered in accordance with the determined current-cycle score;

determining by the processor, for each of the data records in the set for which a score was determined in a prior cycle, whether the current-cycle score is greater than a most recently determined prior cycle score;

generating by the processor a report including at least some of the data records having no score determined in any prior cycle and the data records assigned scores in one or more prior cycles determined to have a greater score than the most-recently determined prior score for the same data record;

determining by the processor a velocity of change in group to which each data record is assigned;

based at least in part on the determined velocity of change in group to which a data record is assigned, selecting by the processor a plurality of the data records for forwarding to an operation system, wherein a positive velocity of change in group renders a data record more likely to be selected for forwarding, and a negative velocity of change in group renders a data record less likely to be selected for forwarding; and

forwarding, via bulk download using xml documents, to the operation system, the data records selected for forwarding.

12. The computer-implemented method of claim 11, wherein the selected previously-reviewed data records comprise data records not forwarded to the operation system and having selected aging intervals.

13. The computer-implemented method of claim 12, wherein the selected aging intervals include a first aging interval of between 10 and 45 days and a second aging interval of at least 180 days.

14. The computer-implemented method of claim 11, further comprising storing in the one or more data storage devices data relating to the previously reviewed data records not having a greater score than a most-recently determined prior score for the same data record.

15. The computer-implemented method of claim 11, wherein the predictive model comprises a plurality of structured data values and coefficients associated with each of the structured data values.

16. The computer-implemented method of claim 11, wherein the data records comprise data records indicative of claims under property insurance policies.

17. A non-transitory computer-readable storage medium, the computer-readable storage medium having processor-executable instructions stored, which instructions, when executed by a processor, cause the processor to:

on a recurring and dynamic basis on a cycle, in each of a plurality of cycles, access from one or more data storage devices stored data relating to data records to determine a set of data records for review, the selected data records including data records having no score determined in any prior cycle and selected data records assigned scores in one or more prior cycles, the selected data records assigned scores in one or more prior cycles comprising data records having selected aging intervals, the stored data comprising structured data and unstructured data, the unstructured data comprising data indicative of communications between two or more persons;

process using data mining the data indicative of the unstructured data to extract words and phrases from the unstructured data of the data records of the set of data records, comprising using text parsing logic to identify sentences, and tagging identified words to be associated with categories and subcategories;

apply a predictive model for determining a current-cycle score indicative of suitability of data records for forwarding to an operation system, for each of the data records of the set, based on review of the structured data and words and phrases extracted from the unstructured data, to the structured data and the extracted words and phrases from the unstructured data to determine a score associated with each of the data records of the set, the predictive model comprising data values and coefficients associated with structured data certain data values being negative and indicative of a reduced suitability for forwarding, and certain data values being positive and indicative of an increased suitability for forwarding, and assign each of the data records to one of a plurality of groups of equal numbers ordered in accordance with determined current-cycle scores;

determine, for each of the data records having a score determined in one or more prior cycles, whether the current-cycle score is greater than a most recently determined prior cycle score;

generate a report including at least some of the data records having no score determined in any prior cycle and the data records assigned scores in one or more prior cycles determined to have a greater score than the most-recently determined prior score for the same data record;

determine a velocity of change in group to which each data record is assigned;

based at least in part on the determined velocity of change in group to which a data record is assigned, select a plurality of the data records for forwarding to the operation system, wherein a positive velocity of change in group renders a data record more likely to be selected for forwarding, and a negative velocity of change in group renders a data record less likely to be selected for forwarding; and

forward, via bulk download using xml documents, to the operation system, the data records selected for forwarding.

18. The non-transitory computer-readable storage medium of claim 17, wherein the instructions further cause the processor to select data records having no score determined in any prior cycle for inclusion in the report based on the determined score.

19. The non-transitory computer-readable storage medium of claim 17, wherein the instructions further cause the processor to transmit the report to the operation server for further review.