Validation of data in a database record using a reinforcement learning algorithm

Abstract

Embodiments are directed to methods and systems for validating data in database records. Each record can comprise a record of a service provided to a consumer and can include a code related to the service. Goals can be defined for the validating of the data assigned to the one or more fields. Each goal can be related to a different possible result of the validating of the data assigned to the one or more fields. A weight can be defined for each goal. Records can be selected using a reinforcement learning algorithm and based on an expected return value for further processing of the selected one or more records. The expected return value can be based on the weight for each goal and a probability of satisfying each defined goal by further processing of the record. The selected records can then be processed according to one or more workflows.

What is claimed is: 
     
         1 . A method for validating data assigned to one or more fields in each of a plurality of database records, the method comprising:
 maintaining, by a records management and processing system, a plurality of records in a database, each record of the plurality of records comprising a record of a service provided to a consumer by a service provider of a plurality of service providers, the one or more fields of each record comprising a code related to the service and wherein the records management and processing system comprises an intermediary between systems of the plurality of service providers and systems of a plurality of responsible entities;   defining, by the records management and processing system, a plurality of goals for the validating of the data assigned to the one or more fields, each goal related to a different possible result of the validating of the data assigned to the one or more fields;   defining, by the records management and processing system, a weight for each goal of the plurality of goals for the validating of the data assigned to the one or more fields;   selecting, by the records management and processing system, one or more records of the plurality of records using a reinforcement learning algorithm and based on an expected return value for further processing of the selected one or more records, wherein the expected return value is based on the weight for each goal of the plurality of goals for the validating of the data assigned to the one or more fields and a probability of satisfying each defined goal by further processing of the record; and   processing, by the records management and processing system, the selected one or more records according to one or more workflows executed by the records management and processing system.   
     
     
         2 . The method of  claim 1 , wherein the reinforcement learning algorithm comprises a multi-arm bandit algorithm. 
     
     
         3 . The method of  claim 2 , wherein selecting the one or more one or more records of the plurality of records based on the expected return value for further processing of the selected one or more records comprises maximizing a total expected value for further processing of the selected records based on the plurality of goals. 
     
     
         4 . The method of  claim 3 , wherein the expected return value for a record comprises a sum of the probabilities of satisfying each defined goal by further processing of the record weighted by the defined weight for the defined goal. 
     
     
         5 . The method of  claim 4 , wherein the goals comprise two or more of a goal directed to compliance with a set of predefined requirements for the one or more fields in each record, a goal directed to denials of records by the plurality of responsible entities based on an incorrect value for the one or more fields in a record, and a goal directed to a level of reimbursement revenue from the plurality of responsible entities for the plurality of records. 
     
     
         6 . The method of  claim 4 , further comprising determining, by the records management and processing system, a value for gathering additional information by further processing of a record, wherein the expected return value is further based on the determined value for gathering additional information by further processing of a record and wherein the determined value for gathering additional information by further processing of a record is inversely proportional to a number of times records with a same type have been previously processed. 
     
     
         7 . The method of  claim 4 , wherein selecting the one or more records of the plurality of records using the reinforcement learning algorithm and based on the expected return value for further processing of the selected one or more records comprises selecting a first set of one or more records from a first service provider and wherein the method further comprises:
 selecting, by the records management and processing system, a second set of one or more records from a second service provider using the reinforcement learning algorithm and based on an expected return value for further processing of the selected second set of one or more records; and   determining, by the records management and processing system, a final score for the selected first set of one or more records and the selected second set of one or more records together, wherein the final score comprises a sum of the expected value for the selected first set of one or more records weighted by a confidence factor for the first service provider and the expected value for the selected second set of one or more records weighted by a confidence factor for the second service provider and wherein processing the selected first set of one or more records and the selected second set of one or more records is further based on the total score.   
     
     
         8 . The method of  claim 1 , wherein processing the selected one or more records according to one or more workflows comprises initiating an audit of each of the selected one or more records. 
     
     
         9 . The method of  claim 1 , wherein initiating the audit of each of the selected one or more records comprises one or more of selecting an agent to conduct the audit, prioritizing the selected one or more records in a work queue of an agent, updating a set of audit records based on results of performing the audit, and updating the weight for one or more goals of the plurality of goals based on results of performing the audit. 
     
     
         10 . A system comprising:
 a processor; and   a memory coupled with and readable by the processor and storing therein a set of instructions which, when executed by the processor, causes the processor to validate data assigned to one or more fields in each of a plurality of database records by:
 maintaining a plurality of records in a database, each record of the plurality of records comprising a record of a service provided to a consumer by a service provider of a plurality of service providers, the one or more fields of each record comprising a code related to the service and wherein the records management and processing system comprises an intermediary between systems of the plurality of service providers and systems of a plurality of responsible entities; 
 defining a plurality of goals for the validating of the data assigned to the one or more fields, each goal related to a different possible result of the validating of the data assigned to the one or more fields; 
 defining a weight for each goal of the plurality of goals for the validating of the data assigned to the one or more fields; 
 selecting one or more records of the plurality of records using a reinforcement learning algorithm and based on an expected return value for further processing of the selected one or more records, wherein the expected return value is based on the weight for each goal of the plurality of goals for the validating of the data assigned to the one or more fields and a probability of satisfying each defined goal by further processing of the record; and 
 processing the selected one or more records according to one or more workflows executed by the records management and processing system. 
   
     
     
         11 . The system of  claim 10 , wherein the reinforcement learning algorithm comprises a multi-arm bandit algorithm. 
     
     
         12 . The system of  claim 11 , wherein selecting the one or more one or more records of the plurality of records based on the expected return value for further processing of the selected one or more records comprises maximizing a total expected value for further processing of the selected records based on the plurality of goals. 
     
     
         13 . The system of  claim 12 , wherein the expected return value for a record comprises a sum of the probabilities of satisfying each defined goal by further processing of the record weighted by the defined weight for the defined goal. 
     
     
         14 . The system of  claim 13 , wherein the goals comprise two or more of a goal directed to compliance with a set of predefined requirements for the one or more fields in each record, a goal directed to denials of records by the plurality of responsible entities based on an incorrect value for the one or more fields in a record, and a goal directed to a level of reimbursement revenue from the plurality of responsible entities for the plurality of records. 
     
     
         15 . The system of  claim 13 , wherein the instructions further cause the processor to determine a value for gathering additional information by further processing of a record, wherein the expected return value is further based on the determined value for gathering additional information by further processing of a record and wherein the determined value for gathering additional information by further processing of a record is inversely proportional to a number of times records with a same type have been previously processed. 
     
     
         16 . The system of  claim 13 , wherein selecting the one or more records of the plurality of records using the reinforcement learning algorithm and based on the expected return value for further processing of the selected one or more records comprises selecting a first set of one or more records from a first service provider and wherein the instructions further cause the processor to:
 select a second set of one or more records from a second service provider using the reinforcement learning algorithm and based on an expected return value for further processing of the selected second set of one or more records; and   determine a final score for the selected first set of one or more records and the selected second set of one or more records together, wherein the final score comprises a sum of the expected value for the selected first set of one or more records weighted by a confidence factor for the first service provider and the expected value for the selected second set of one or more records weighted by a confidence factor for the second service provider and wherein processing the selected first set of one or more records and the selected second set of one or more records is further based on the total score.   
     
     
         17 . The system of  claim 10 , wherein processing the selected one or more records according to one or more workflows comprises initiating an audit of each of the selected one or more records. 
     
     
         18 . The system of  claim 10 , wherein initiating the audit of each of the selected one or more records comprises one or more of selecting an agent to conduct the audit, prioritizing the selected one or more records in a work queue of an agent, updating a set of audit records based on results of performing the audit, and updating the weight for one or more goals of the plurality of goals based on results of performing the audit. 
     
     
         19 . A non-transitory, computer-readable medium comprising a set of instructions stored therein which, when executed by a processor, causes the processor to validate data assigned to one or more fields in each of a plurality of database records by:
 maintaining a plurality of records in a database, each record of the plurality of records comprising a record of a service provided to a consumer by a service provider of a plurality of service providers, the one or more fields of each record comprising a code related to the service and wherein the records management and processing system comprises an intermediary between systems of the plurality of service providers and systems of a plurality of responsible entities;   defining a plurality of goals for the validating of the data assigned to the one or more fields, each goal related to a different possible result of the validating of the data assigned to the one or more fields;   defining a weight for each goal of the plurality of goals for the validating of the data assigned to the one or more fields;   selecting one or more records of the plurality of records using a reinforcement learning algorithm and based on an expected return value for further processing of the selected one or more records, wherein the expected return value is based on the weight for each goal of the plurality of goals for the validating of the data assigned to the one or more fields and a probability of satisfying each defined goal by further processing of the record; and   processing the selected one or more records according to one or more workflows executed by the records management and processing system.   
     
     
         20 . The non-transitory, computer-readable medium of  claim 19 , wherein the reinforcement learning algorithm comprises a multi-arm bandit algorithm, wherein selecting the one or more one or more records of the plurality of records based on the expected return value for further processing of the selected one or more records comprises maximizing a total expected value for further processing of the selected records based on the plurality of goals, wherein the expected return value for a record comprises a sum of the probabilities of satisfying each defined goal by further processing of the record weighted by the defined weight for the defined goal, and wherein the goals comprise two or more of a goal directed to compliance with a set of predefined requirements for the one or more fields in each record, a goal directed to denials of records by the plurality of responsible entities based on an incorrect value for the one or more fields in a record, and a goal directed to a level of reimbursement revenue from the plurality of responsible entities for the plurality of records.