WO2008098167A2

WO2008098167A2 - Robot and web-based method for affiliation verification

Info

Publication number: WO2008098167A2
Application number: PCT/US2008/053415
Authority: WO
Inventors: Thomas M. Digiacinto; Lorraine K. Bauman; Karen Swenson; Dan Adams
Original assignee: Ims Software Services, Ltd.
Priority date: 2007-02-08
Filing date: 2008-02-08
Publication date: 2008-08-14
Also published as: WO2008098167A3

Abstract

A robotic tool confirms information on organization-to-organization and individual-to-organization relationships recorded in a database. The robotic tool conducts fully automated open Internet searches to improve the accuracy and quality of organization and professional affiliations recorded in the database. All professional and organizational affiliations in the database can continually verified and confirmed by fully automated search and result-scoring of public domain content on the Internet.

Description

ROBOT AND WEB-BASED METHOD FOR AFFILIATION VERIFICATION

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority to U.S. Provisional Patent Application Serial No. 60/888,891, filed February 8, 2007, which is incorporated herein by reference in its entirety.

BACKGROUND The present invention relates to the management of information or data on individuals, service providers, organizations and entities involved in complex endeavors, for example, in the health care industry.

The delivery of health care to customers is a complex endeavor involving many individuals (patients, physicians, pharmacists, etc.) and organizations (e.g., hospitals, insurance companies, R&D groups, drug manufacturers, etc.).

Investigations in the health care field (e.g., demographic or epidemiological studies, pharmaceutical testing or market research, etc.), which may extend over a substantial period of time, are based on collected information attributable to particular individuals at various levels in many organizations and entities. For the investigations, it is often necessary to understand organization-to-organization and individual-to-organization relationships.

Electronic databases with data on organization profiles, relationships and professional affiliations are useful to the investigators. For example, IMS Health, Inc., assignee of the present invention, maintains a comprehensive and up-to-date organization profiles, relationships and professional affiliations database (e.g., Healthcare Organization Services (HCOS) database). The HCOS information can be integrated with other health care transactional activity information (e.g., IMS transactional DDD and Xponent data). The HCOS database provides timely reference information (i.e., organization profiles, relationships and professional affiliations) on hospitals, medical group practices, long-term care facilities, outpatient surgery centers, diagnostic imaging centers and home-health agencies. The HCOS database includes comprehensive data on 200,000-plus integrated DDD outlet links (shipping locations) and more than 2 million professional affiliations that can be integrated with Xponent for prescribing transactions or with PlanTrak for managed care prescribing activity. Consisting of more than 450,000 organizations in approximately 150 different classes-of-trade (COT), HCOS data provides details about organizations, such as hospitals and outpatient centers, and their purchasing/contracting relationships to other organizations such as corporate parents, group purchasing organizations (GPO), pharmacy providers and distributors. Hospitals include acute care, psychiatric and rehabilitation facilities; outpatient centers include medical groups that focus on oncology, family practice and rheumatology as well as other areas of specialization.

In addition to provider affiliations, HCOS data includes names and titles of more than 100,000 administrative and clinical decision makers at hospitals, nursing homes, corporate parents and GPOs. HCOS data is available with all industry- standard identifiers. For organizations, these industry-standard identifiers include DDD, DEA, HIN, MPN and PHS. For professionals, these industry-standard identifiers include IMS ID, DEA, ME number and UPIN. This comprehensive cross- reference supports clients' data-integration efforts and linkage to IMS transactional DDD and Xponent sales data. Consideration is now being given to improving information databases that are used for marketing of products (e.g., pharmaceutical products). Information on organization profiles, relationships and professional affiliations of entities in the market environment is of interest. In particular, attention is being paid to product marketing in an environment in which several interacting market variables can influence results.

SUMMARY OF THE INVENTION

A tool and a web-based method for confirming information recorded in a database are provided. The tool and method, which are based on open Internet searches, can be used for improving the accuracy and quality of organization and professional affiliations recorded in a database on organization-to-organization and individual-to-organization relationships. The tool and method enable all professional and organizational affiliations in a database to be continually verified or confirmed by fully automated search- and result scoring of returned Internet, public domain content.

BRIEF DESCRIPTION OF THE DRAWINGS

Further features of the invention, its nature, and various advantages will be more apparent from the following detailed description of the preferred embodiments and the accompanying drawings in which:

FIG. 1 is a schematic illustration of exemplary prescriber-affiliation identification VBOT processes for updating affiliation databases, in accordance with the principles of the present invention.

FIG. 2 is a flow diagram which illustrates an exemplary process flow for verifying prescriber affiliations, in accordance with the principles of the present invention.

FIG. 3 is an illustration of the results of a test run of the exemplary process of FIG. 2, in accordance with the principles of the present invention.

DESCRIPTION OF THE INVENTION

A tool and a web-based method for verifying and confirming information recorded in a database are provided. The tool and method, which are based on open Internet searches, can be used for improving the accuracy and quality of organization and professional affiliations recorded in a database on organization-to-organization and individual-to-organization relationships. The tool and method enable all professional and organizational affiliations in a database to be continually verified or confirmed by fully automated search and result-scoring of returned Internet, public domain content.

Data about an individual (e.g., a healthcare professional), organization and affiliation is run through specialized string algorithms to determine the most efficient criteria to establish the "scope" of an open Internet search. Once the scope sites have been located, additional algorithms may be applied against the located data content as part of a "refined" search to isolate the necessary information to confirm or verify the individual's affiliation. Further, a precision or accuracy score may be applied to the affiliation based on the number and relevance of the refined keywords that are located in the search.

For convenience in description, both the robot and the method may be collectively referred to herein as the "robot" (VBOT), where appropriate. Further, for convenience, the operation of VBOT is described herein with reference to the exemplary HCOS database, which includes organization relationships and physician affiliations.

VBOT is a web-based software process designed to enable all professional and organizational affiliations to be continually confirmed using fully automated search and result-scoring of returned Internet, public domain content. The process, which can be run continuously (i.e., 24 hours-7 days a week) on a dedicated server, searches open Internet search engines to identify and/or confirm organizational-professional affiliations. The number of affiliations in a database such as HCOS is currently 2.5 million and increasing. VBOT in conjunction with, suitable metrics enables "active" validation of the large number of affiliations in the database on a continuous cycle.

The relationships (e.g., physician affiliations) in HCOS are established using a combination of primary research and secondary research, and by data integration with market intelligence core data, claims data sources and the DEA database. The most accurate affiliation information, however, is obtained through manual research and review. This manual research and review process can result in high precision, but is labor intensive and expensive. The inventive VBOT advantageously provides automated processes to confirm existing physician affiliations, thereby eliminating the need for manual review. Also, VBOT can identify questionable physician affiliations as well as potential new physician affiliations, both of which can be addressed by primary research follow-up (e.g., by database builders). In addition to contributing to the precision of the HCOS physician affiliation data, VBOT meets client requests for greater affiliation accuracy.

VBOT may confirm all physician affiliations within weeks, using a 24-7 automated process. In an exemplary implementation, assuming 2 instances, VBOT can confirm 50,000 affiliations per week or the entire HCOS affiliations database every 50 weeks. Assuming 4 instances, VBOT can confirm 100,000 affiliations per week or the entire HCOS database every 25 weeks. With these exemplary implementations, it is possible to provide database clients with useful daily, weekly or monthly updates to affiliations and an HCOS affiliations database that is fully refreshed semi-annually or annually.

The advantages of the automated VBOT processing can be appreciated upon an understanding of the prior art construction of the HCOS database, in which the current 2.5 million affiliations are derived by automated matching of prescriber addresses (maintained by assignee) against HCOS organizational addresses, and by manual research and review processes. These affiliations represent one million prescribers at almost 250,000 organizations. Table I below shows the quantity of affiliations that are processed presently using non-automated prior art techniques (e.g., primary research, web searching and claims data) on an annual basis. TABLE I

^'MotilfϊeφgC , Modified Q4 Averag

Primary Research - Active 12,516 18,877 23,291 23,648 19,583

Primary Research - Deactive 20,451 39,083 27,099 22,410 27,261

Bulk Fast Link 11 ,647 16,751 20,122 36,966 21 ,372

Fast Link 10,983 15,591 ^r9?Ϊ84^~ 22,802 14,640

Claims Data 8,562 1 ,261 5,883 1 ,580 4,322

Total 64,159 91 ,563 85,579 107,406 87,177

(-90,

Using the prior art techniques (e.g., phone calls and manual directory searches), it is possible to validate and/or create an average of 90,000 affiliations per quarter. However, with 2.5 million affiliations, it would take seven (7) years (360,000 per year) to source-verify the full affiliation file in the HCOS database.

In contrast to slow prior art techniques, preliminary VBOT tests indicate that a fully optimized VBOT will provide VBOT validation of a minimum of 25,000 affiliations per week per VBOT instance or run. VBOT can be implemented so that multiple instances can run concurrently using different affiliation subsets. For example, "high decile" prescribers with outpatient affiliations may be run in one instance against Search Engine A while simultaneously "low decile" prescribers or physician extenders may be run against Search Engine B, etc.

As previously noted, assuming two runs, VBOT can confirm 50,000 affiliations per week or the entire affiliation database every 50 weeks. Assuming four runs, VBOT can confirm 100,000 affiliations per week or the entire database every 25 weeks. These VBOT run rates allow the database operators to provide clients with daily, weekly or monthly updates to affiliations and an HCOS affiliations database that is fully refreshed semi-annually or annually.

For client projects, VBOT can process 3600 professionals per day, per instance or run. This allows client files to be VBOT "scrubbed" for precision and accuracy, as a first step in all client projects.

An exemplary VBOT, which is designed for HCOS database updating, may include automated processes to validate the outpatient affiliations volume in an existing HCOS database (e.g., 1 million affiliations for 650,000 professionals) for accuracy every 5 weeks (assuming four VBOT instances or runs). Further, the exemplary VBOT may include automated processes to assist in locating affiliations for prescribers who do not have an outpatient affiliation (e.g., 1.25 million non- physician extenders). Additionally, the exemplary VBOT may include automated processes to assist in locating affiliations for medical groups that do not currently meet the requirement of 2 or more linked prescribers (-38,000 organizations).

For verification of Prescriber Affiliation (PA), VBOT may be configured to operate in two modes or steps:

1. Verify - Take existing outpatient affiliations within HCOS and search the web for information that confirms the affiliation. Affiliations may be tagged with a VBOT validation score based on the number and quality of the search results.

2. Affiliate - Search for potential affiliations for prescribers without an outpatient affiliation (excluding IPPs) to known HCOS outpatient facilities. For verification of Organization Affiliation Search (OA), VBOT may be configured to take outpatient facilities that have no (zero) or one (1) affiliated prescriber, and search for potential affiliations on the Internet.

FIG. 1 shows exemplary prescriber-organization affiliation identification VBOT process 100 for updating affiliation databases (e.g., Prescriber/affiliation database 110). Prescriber/affiliation database 110 may have been constructed using existing processes, for example, using human researchers 130, to confirm and document prescriber affiliations. Data entry of new prescriber affiliations may be controlled or supervised by Mil researchers 130.

In process 100, search string extracts 120 (e.g., prescriber name and characteristic: "Jonathan Smith + NJ") may be obtained from Prescriber/affiliation database 110. In the case of a new prescriber (i.e., a prescriber with affiliations not listed in Prescriber/affiliation database 110), a new extract is defined independent of database 110. The string extracts are defined and assembled in a Daily Search String Database 140. VBOT instances or runs (e.g., daily) involve searching the Internet using search terms equal to string extracts 120 from Database 140. Any suitable Internet search engine 150 (including publicly available search engines such as Google and Ask) may be used for conducting automated searches. In each VBOT instance or run the Internet is automatically searched for search terms set equal to search strings in a list selected from Daily Search String Database 140. Search results 160, which identify affiliations for a prescriber, are analyzed and assigned scores 170. According to the score values (e.g., positive score values 180), existing affiliations for a prescriber listed in Prescriber/affiliation database 110 are validated (190). In the case of a new prescriber, whose affiliations are not previously listed in database 110, the new prescriber affiliations discovered by VBOT searching may be recorded in a New Prescriber- Affiliations database 195. The new prescriber- affiliation data may be made available to Mil researcher 130 for independent verification (e.g., using conventional methods). Preferably, Mil researcher 130 confirms and documents the new prescriber-affiliation data before it is entered in Prescriber/affiliation database 110.

Exemplary process 100 may have a configurable mechanism for selecting target segment(s) (i.e., search string extracts 120/140) from database 110 data records for VBOT processing based on suitable data elements, for example, Prescriber profession, Prescriber specialty sub-group or specialty group, Prescriber Decile, Target Prescriber List (including client-supplied target list), State, COT Classification, COT Facility Type or select Class of Trade(s), DDD Subcat, Target DDD or IMS ORG ID List (including client-supplied target list), and/or Affiliation Approval Level and/or modification date of the affiliation. The exemplary VBOT may have multiple levels of VBOT Searching, for example, a two-level Scope Search and Refined Search.

The Scope Search may, for example, include a search on 3-6 generalized factors, such as last name, profession abbreviation, specialty sub-group, city and/or state to identify the initial scope of data to be returned. The Scope Search may provide a selected number of search returns (e.g., up to 10 web sites found) and the URL ofthe site(s). The Refined Search may further search the search returns (e.g., the top 10 web pages returned by the Scope Search) for more specific keywords, for example, first name, middle name, address, and organization name.

Further, the exemplary VBOT may have a VBOT Score, which is based on the number of sites and the specificity of the keyword search. Alternatively, the VBOT score may be assigned based on business rules and data analysis. The VBOT score may be used as a metric to rate existing affiliations, mark existing affiliations as suspicious enough to require research review, and/or to suggest/create new affiliations.

FIG. 2 is a flow diagram which illustrates an exemplary process flow 200 for verifying prescriber affiliations in a prescriber-affiliations database ("Phoenix"), in accordance with the principles of the present invention.

Process flow 200 includes one or more subprocesses 210-290, which are as follows:

210: Determine Data Services (i.e., select target data segment such as decile

10 primary care prescribers to validate and select an appropriate search engine to use).

220: Load Scope Data for selected Segment (i.e., generic criteria to locate top 10 web sites from search engine, such as first name, last name, city, state).

230: Load Refined Data for selected Segment (i.e., specific affiliation information under review, such as business name and provider specialty).

240: Execute VBOT engine. 250: Score results by applying a scoring algorithm.

260: Apply score to existing affiliation in HCOS Affiliations database.

270: Optionally, conduct manual research based on the results based on the score.

280: Optionally, perform automated tasks based on the score (e.g., validate or update existing prescriber-affiliation data records).

290: Load new or updated data records in database as appropriate or desired.

FIG. 3 shows results of a test run of exemplary process 200 for verification of

Outpatient Prescriber Affiliations. In the test run, outpatient affiliations for all prescribers with a Decile of 10 (corresponding to 10,806 affiliations) were investigated. The options selected for subprocesses 210-240 in the test run are shown in TABLE II.

TABLE II

The foregoing merely illustrates the principles of the invention. Various modifications and alterations to the described embodiments will be apparent to those skilled in the art in view of the teachings herein. It will thus be appreciated that those skilled in the art will be able to devise numerous techniques which, although not explicitly described herein, embody the principles of the invention and are thus within the spirit and scope of the invention.

It will be understood that in accordance with the present invention, the robot and the method described herein may be implemented using any suitable combination of hardware and software. The software (i.e., instructions) for implementing and operating the aforementioned robot and the method can be provided on computer- readable media, which can include, without limitation, firmware, memory, storage devices, microcontrollers, microprocessors, integrated circuits, ASICs, on-line downloadable media, and other available media.

Claims

1. An automatic computerized method for validating information recorded in a database, the method comprising: automatically assembling a search string extracts list of one or more search string extracts from the information recorded in the database; running automated open Internet searches using the one or more search string extracts in the list as search terms; automatically evaluating and accordingly assigning quality scores to the automated open Internet search results; and automatically validating the search string extract information recorded in the database using the automated open Internet search results according to the quality scores assigned thereto.

2. The method of claim 1 , wherein the database is an organization profiles, relationships and professional affiliations database of organizations and individuals in the health care industry.

3. The method of claim 1, further comprising applying string algorithms to data about an individual, organization and affiliation to determine a scope of the open Internet searches, wherein the Internet search results comprise scope sites.

4. The method of claim 3, further comprising searching the scope search results data content using refined search terms or keywords to isolate information that confirms or verifies an individual's affiliation.

5. The method of claim 4, wherein the quality scores comprise a precision or accuracy score applied to the individual's affiliation based on the number and relevance of the refined keywords that are located in the search.

6. The method of claim 1 , wherein the database is an prescriber- outpatient affiliations database, wherein automatically assembling a search string extracts list of one or more search string extracts from the information recorded in the database and running automated open Internet searches comprises searching for potential affiliations for prescribers without an outpatient affiliation to known outpatient facilities.

7. The method of claim 1, wherein the database is a prescriber-outpatient facilities affiliations database, and wherein automatically validating the search string extract information recorded in the database using the automated open Internet search results according to the quality scores assigned comprises confirming existing affiliations in the database.

8. The method of claim 1, wherein the database is a prescriber-affiliations database, and wherein the method comprises determining the data services to be used including selecting target data segments of the database to be validated and selecting an appropriate search engine.

9. The method of claim 8, wherein the method comprises, for the selected segments, establishing generic criteria including name and state for a scope search by the search engine.

10. The method of claim 9, wherein the method comprises, for the selected segments, further establishing specific criteria including affiliation information under review for a refined search by the search engine.

11. Computer-readable media comprising instructions that can be executed to: automatically assemble a search string extracts list of one or more search string extracts from the information recorded in the database; run automated open Internet searches using the one or more search string extracts in the list as search terms; automatically evaluate and accordingly assign quality scores to the automated open Internet search results; and automatically validate the search string extract information recorded in the database using the automated open Internet search results according to the quality scores assigned thereto.

12. The computer readable media of claim 11 , wherein the instructions comprise instructions that are executable to apply string algorithms to data about an individual, organization and affiliation to determine a scope of the open Internet searches, wherein the Internet search results comprise scope sites.

13. The computer readable media of claim 12, further comprising instructions that are executable for searching the scope sites data content using refined search terms or keywords to isolate information that confirms or verifies an individual's affiliation.

14. The computer readable media of claim 13, further comprising instructions that are executable to generate a precision or accuracy score applied to the individual's affiliation based on the number and relevance of the refined keywords in the scope sites data content.

15. The computer readable media of claim 13 , wherein the instructions comprise instructions that are executable to determine the data services to be used including selecting target data segments of the database to be validated.

16. An automatic tool for validating information recorded in a database, the tool comprising: an assembler configured to assemble a search string extracts list of one or more search string extracts from the information recorded in the database; an search engine configured to run automated open Internet searches using the one or more search string extracts in the list as search terms; a results evaluator configured to apply scoring algorithms to evaluate and assign assigning quality scores to the automated open Internet search results; and an automated validator configured to validate the search string extract information recorded in the database using the automated open Internet search results according to the quality scores assigned thereto.

17. The automatic tool of claim 16 further comprising algorithms to score an individual's affiliation based on the number and relevance of keywords that are located in the search.