Patent Application Titled “Machine Learning Model For Predicting Health Plans Based On Missing Input Data” Published Online (USPTO 20230394588): Patent Application

2023 DEC 27 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- According to news reporting originating from Washington, D.C., by NewsRx journalists, a patent application by the inventors DELANO, III, Raymond G. (Leawood, KS, US); JENSEN, Julie Ann (Olathe, KS, US); POTEET, III, James L. (Overland Park, KS, US), filed on August 18, 2023, was made available online on December 7, 2023.

No assignee for this patent application has been made.

Reporters obtained the following quote from the background information supplied by the inventors: “Health care providers are dependent on accurate insurance information to ensure medical claims are correctly generated, matched and processed with medical plan data contained in electronic databases. Often patient insurance eligibility and health care claims are generated incorrectly and/or transmitted to the wrong payer or health plan, resulting in errors or denied claims. This situation may happen if the health plan on the insurance card cannot be matched to a health plan in the electronic health record system (EHR). Or, perhaps the health plan was built in the EHR using a slightly different name or is abbreviated or confused with or similarly named plan. Similar situations may arise if the patient is from a different state or locality and the provider organization has not built a particular health plan in the EHR. Sometimes, the health insurance card is scanned correctly but still manually matched to the wrong health plan and/or the electronic routing is configured incorrectly. In each of these scenarios, the electronic transactions are performed incorrectly and thus requiring corrective actions and repeating the transactions, which uses more computing system resources and processing time.”

In addition to obtaining background information on this patent application, NewsRx editors also obtained the inventors’ summary information for this patent application: “This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

“Embodiments of the present system relate to improving the technical field of systematically predicting payer/health plan combinations of data records with a machine learning model for situations where an insurance card is not presented or the card does not contain requested data (e.g., some identifiers are missing). Thus, the machine learning model makes predictions when initial input data is missing. In some embodiments, the system relates to utilizing machine learning to verify payers and/or health plans. More particularly, the present invention utilizes HIPAA transactions to train a machine learning model to intelligently link payers and/or health plans to specific employers. Initially, transaction data is received from electronic data interchange (EDI) insurance transactions. The transaction data comprises data corresponding to a plurality of employers, a plurality of payers, and a plurality of health plans provided by the plurality of payers. A machine learning model is trained with the transaction data to build a mapping of the plurality of employers, the plurality of payers contracted with each employer of the plurality of employers, and the plurality of health plans provided by the plurality of payers for each employer of the plurality of employers. The machine learning model is configured to generate and display on a display, a predicted set of one or more payer-health plan combinations that are valid with an employer identifier that corresponds to an inputted employer and work location data.”

The claims supplied by the inventors are:

“1. A non-transitory computer-readable medium storing computer-executable instructions that, when executed by a computer including a processor, cause the computer to perform functions comprising: training a machine learning model to predict a payer-health plan combination in response to missing data by: receiving transaction data from electronic data interchange (EDI) insurance transactions, the transaction data comprising data corresponding to a plurality of employers, a plurality of payers contracted with one or more employers of the plurality of employers, and a plurality of health plans provided by the plurality of payers for one or more employers of the plurality of employers; extracting features from the received transaction data that identify combinations of employer identifier, payer identifier, and health plan identifier that are valid combinations of health plans; and training the machine learning model to predict a payer-health plan combination by inputting the valid combinations of employer identifier, payer identifier, and health plan identifier to the machine learning model; in response to a request to identify a valid combination, obtaining employer and work location data associated with a patient who does not present an insurance card; inputting the employer and work location data into the machine learning model; and generating, by the machine learning model, and displaying on a display, a predicted set of one or more payer-health plan combinations that are valid with an employer identifier that corresponds to the employer and work location data that is inputted.

“2. The non-transitory computer-readable medium of claim 1, further comprising instructions that, when executed by at least the processor, cause the processor to: provide a user interface that displays fields for an identification of the payer, an identification of the health plan of the payer, and an identification of the employer.

“3. The non-transitory computer-readable medium of claim 2, further comprising instructions that, when executed by at least the processor, cause the processor to: automatically extract, at a card scanning service, scan data into appropriate fields of the user interface, the scan data corresponding to an insurance card and comprising the identification of a payer and the identification of a health plan of the payer.

“4. The non-transitory computer-readable medium of claim 3, further comprising instructions that, when executed by at least the processor, cause the processor to: enable a user, via the user interface, to review or revise the extracted scan data prior to saving the extracted scan data in an electronic health record of a patient.

“5. The non-transitory computer-readable medium of claim 3, wherein the extracted scan data comprises the identification of the payer, the identification of the health plan of the payer, and the identification of the employer.

“6. The non-transitory computer-readable medium of claim 3, further comprising instructions that, when executed by at least the processor, cause the processor to: display, at the user interface, a view of a front image capture and a view of a back image capture of the insurance card.

“7. The non-transitory computer-readable medium of claim 1, further comprising instructions that, when executed by at least the processor, cause the processor to: display, on a user interface, the employer and work location data imported from a registration system.

“8. The non-transitory computer-readable medium of claim 1, further comprising instructions that, when executed by at least the processor, cause the processor to: display, on a user interface, a filtered list of the valid payer/health plan combinations based on the inputted employer and work location data.

“9. The non-transitory computer-readable medium of claim 8, further comprising instructions that, when executed by at least the processor, cause the processor to: enable, via the user interface, a user to select a payer/health plan combination for a patient.

“10. A method for predicting a payer/health plan combination for a patient comprising: training a machine learning model to predict valid combinations of employer-payer-health plan in response to one or more missing identifiers by: receiving transaction data from electronic data interchange (EDI) insurance transactions, the transaction data comprising data corresponding to valid combinations of a plurality of employers, a plurality of payers contracted with each employer of the plurality of employers, and a plurality of health plans provided by the plurality of payers for each employer of the plurality of employers; extracting features from the received transaction data that identify combinations of employer identifier, payer identifier, and health plan identifier that are valid combinations; and training a machine learning model to predict valid combinations for a patient by inputting the valid combinations of employer identifier, payer identifier, and health plan identifier to the machine learning model; in response to a request to identify a valid combination based on at least one missing identifier, obtaining one known identifier corresponding to an employer name, a payer name, or a health plan name, and obtaining work location data associated with a patient wherein the other identifiers are missing; inputting the one known identifier and the work location data into the machine learning model; and generating, by the machine learning model, and displaying on a user interface, a predicted set of one or more valid combinations of employer-payer-health plans that correspond to the inputted one known identifier and the work location information, wherein each combination from the predicted set is selectable via the user interface.

“11. The method of claim 10, further comprising: generating claim data for a patient encounter based on a selected valid combination from the user interface; and transmitting the generated claim data as a transaction, via network communication or other communication channel, to a selected third party system for processing.

“12. The method of claim 10, wherein the method further comprises: utilizing the machine learning model to verify that a selected combination from the predicted set is mapped in accordance with the mapping.

“13. The method of claim 10, wherein obtaining the one known identifier comprises: receiving an identification of an employer name and a work location.

“14. The method of claim 13, wherein the method further comprises: predicting, by the machine learning model, the predicted set of one or more valid combinations including one or more payer/health plan combinations based on the identification of the employer and the work location.

“15. The method of claim 14, wherein the method further comprises: enabling a user, via the user interface, to select a payer/health plan combination of the one or more payer/health plan combinations.

“16. The method of claim 15, wherein the method further comprises: saving the selected payer/health plan combination in an electronic health record of a patient.

“17. A system comprising: one or more processors; and a non-transitory computer storage media storing computer-useable instructions that, when used by the one or more processors, cause the one or more processors to: train a machine learning model to predict a payer-health plan combination for a patient in response to missing data by: receiving transaction data from electronic data interchange (EDI) insurance transactions, the transaction data comprising data corresponding to a plurality of employers, a plurality of payers contracted with each employer of the plurality of employers, and a plurality of health plans provided by the plurality of payers for each employer of the plurality of employers; extracting features from the received transaction data that identify combinations of employer identifier, payer identifier, and health plan identifier that are valid combinations; and training the machine learning model to predict a payer-health plan combination for a patient by inputting the valid combinations of employer identifier, payer identifier, and health plan identifier to the machine learning model; in response to a request to identify a valid combination when payer and health plan identifiers are missing, obtain employer and work location data associated with a patient who does not present an insurance card; input the employer and work location data into the machine learning model; and generate, by the machine learning model, a predicted set of one or more payer-health plan combinations that are valid with an employer identifier that corresponds to the inputted employer and work location information.

“18. The system of claim 17, wherein the system is further configured to: receive additional transaction data from the EDI insurance transactions, the additional transaction data comprising (i) approvals of claim data containing valid combinations of employer-payer-health plan identifiers and (ii) denials of claim data containing invalid combinations of employer-payer-health plan identifiers; extract a dataset of features from the additional transaction data corresponding to the approvals and the denials; and retrain the machine learning model based on the approvals and denials.

“19. The system of claim 17, wherein the system is further configured to: generate claim data for a patient encounter based on a selected valid combination from the user interface; and transmit the generated claim data as an electronic transaction, via network communication or other communication channel, to a selected third party system for processing.”

There are additional claims. Please visit full patent to read further.

For more information, see this patent application: DELANO, III, Raymond G.; JENSEN, Julie Ann; POTEET, III, James L. Machine Learning Model For Predicting Health Plans Based On Missing Input Data. U.S. Patent Application Number 20230394588, filed August 18, 2023 and posted December 7, 2023. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20230394588)&db=US-PGPUB&type=ids

(Our reports deliver fact-based news of research and discoveries from around the world.)