“Using Machine Learning To Classify Insurance Card Information” in Patent Application Approval Process (USPTO 20230177617): Walmart Apollo LLC

2023 JUN 26 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- A patent application by the inventors Medous, Frederic (San Francisco, CA, US); Smith, Walter R. (Seattle, WA, US), filed on January 30, 2023, was made available online on June 8, 2023, according to news reporting originating from Washington, D.C., by NewsRx correspondents.

This patent application is assigned to Walmart Apollo LLC (Bentonville, Arkansas, United States).

The following quote was obtained by the news editors from the background information supplied by the inventors: “This disclosure relates generally to medical insurance and more specifically to machine learning algorithms applied to insurance cards for identifying and recommending insurance plans.

“In today’s complex medical insurance market, there are a myriad of choices for medical insurance plans for consumers. The various types of medical insurance plans range from health maintenance organizations (HMOs), preferred provider organizations (PPOs), exclusive provider organizations (EPOs), point-of-service (POS) plans, to high-deductible health plans (HDHPs). Evaluating medical insurance plans as a consumer can be overwhelming when trying to balance the benefits of each plan against the costs of each plan. While insurance policies define intricate cost calculations for healthcare providers and healthcare consumers, they are often written with complicated language and terminology, making the policies challenging to decipher and making it more difficult for consumers to comparison shop. As a result, many consumers enroll in insurance plans that are not the most appropriate for their medical needs and financial budget.”

In addition to the background information obtained for this patent application, NewsRx journalists also obtained the inventors’ summary information for this patent application: “The present invention involves a method for applying machine-learning algorithms to identify an insurance plan associated with an insurance card of a user and providing a recommendation to the user based on the user’s insurance plan. An insurance management system receives, from a client device, at least one image of an insurance card associated with the user operating the client device. The at least one image may include, e.g., a front side of the insurance card and a back side of the insurance card. The insurance management system extracts features from the at least one image, such as text information, object information, color information, and shape information. The insurance management system generates a feature vector associated with the at least one image and inputs the feature vector into a machine learning model that has been trained using classifications of insurance cards of other users. The machine learning model outputs a classification, which comprises an association with an insurance plan. The insurance management system stores the insurance plan of the classification as a current selection of an insurance plan of the user. In one embodiment, the insurance management system receives historical medical information that is specific to the user and indicates a current medical condition of the user. The historical medical information comprises one or more of a prescription history, a medical expense history, a medical claims history, a medical procedure history, and a co-payment history of the user. Based in part on the historical medical information, the insurance management system evaluates the current selection of the insurance plan of the user. Based on the evaluation, the insurance management system provides for display to the user in a user interface a recommendation associated with the current selection of the insurance plan of the user. In one embodiment, the recommendation comprises one or more alternative insurance plans for the user for the user to consider as alternatives to the insurance plan of the user associated with the insurance card. In one embodiment, the historical medical information indicates a pharmacy of the user, and the recommendation comprises one or more alternative pharmacies for the user for the user to consider as alternatives to the pharmacy of the user.

“The figures depict various embodiments of the present invention for purposes of illustration only. One skilled in the art will readily recognize from the following discussion that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the invention described herein.”

The claims supplied by the inventors are:

“1. A system comprising: one or more processors; and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform functions comprising: extracting one or more feature values from at least one image of an insurance card of a user, wherein the at least one image is scanned from a mobile device of the user; generating a feature vector associated with the at least one image of the insurance card of the user; reducing, using dimensionality reduction, an amount of data in the feature vector to a reduced set of data; determining, by a machine learning model, a first insurance plan of the user based on machine learning model input data comprising the reduced set of data and not comprising the amount of data that is absent from the reduced set of data; identifying at least one alternative insurance plan for the user; and sending instructions to display a recommendation of the at least one alternative insurance plan on a user interface of the mobile device of the user.

“2. The system of claim 1, wherein the computing instructions, when executed on the one or more processors, further cause the one or more processors to perform a function comprising: training the machine learning model by: using a set of images of insurance cards as training data for the machine learning model, wherein inputs to the machine learning model comprise historical data for one or more classifications associated with one or more types of insurance cards during a historical time period, wherein outputs of the machine learning model comprise at least one insurance plan matching at least one classification of the one or more classifications associated with the one or more types of insurance cards, and wherein each classification of the one or more classifications is associated with a respective insurance plan.

“3. The system of claim 1, wherein determining, by the machine learning model, the first insurance plan comprises: determining, based on a set of feature values of the one or more feature values, a template of a first insurance card that is similar to the insurance card of the user in the at least one image, wherein the template is a geometrical representation of the first insurance card known to an insurance management system and comprises one or more slots in various locations on the template where each slot is associated with a respective type of information; and populating, by the one or more processors, each of a plurality of particular data fields in the template with one or more respective feature values, as extracted, that represent a respective type of feature to fill each of the plurality of particular data fields.

“4. The system of claim 1, wherein the computing instructions, when executed on the one or more processors, further cause the one or more processors to perform functions comprising: storing the first insurance plan, as determined; receiving historical medical information of the user comprising medical conditions of the user; generating a comparison of a plurality of feature values of the first insurance plan of the user to other feature values of one or more alternative insurance plans, wherein the other feature values comprise one or more of: one or more respective benefits associated with the first insurance plan of the user and each alternative insurance plan of the one or more alternative insurance plans; a respective cost associated with the first insurance plan of the user and each alternative insurance plan; or a respective savings associated with each alternative insurance plan relative to the first insurance plan of the user; selecting the at least one alternative insurance plan of the one or more alternative insurance plans based on the comparison; analyzing a medical expense history of the user comprising at least one of: (i) a medical spending history comprising one or more of health spending account contributions, (ii) health spending account usage, (iii) deductible tracking, or (iv) out-of-pocket expenses of the user; determining a financial prediction for the user based on the medical expense history; and providing the recommendation based on the financial prediction, wherein the recommendation comprises a recommended timeframe for medical visits or medical procedures for the user; wherein: the one or more feature values, as extracted, further comprises object information, color information, and shape information.

“5. The system of claim 1, wherein extracting the one or more feature values comprises: identifying text information from the one or more feature values comprising one or more of: a patient name, an insurance carrier name, an insurance plan name, a patient identification number, an issuer identification number, a group number, a coverage effective date, an employer name, a bank identification number (BIN), co-payment information, or a customer service number.

“6. The system of claim 1, wherein identifying the at least one alternative insurance plan comprises: identifying historical medical information comprising one or more of: a prescription history, a medical expense history, a medical claims history, a medical procedure history, a co-payment history of the user, or a pharmacy of the user.

“7. The system of claim 1, wherein the computing instructions, when executed on the one or more processors, further cause the one or more processors to perform functions comprising: forming a positive training set of images of other insurance cards classified as associated with a respective insurance plan or a respective insurance company; providing the positive training set to the machine learning model for training of the machine learning model; forming a negative training set of images of the other insurance cards classified as not associated with the respective insurance plan or the respective insurance company; providing the negative training set to the machine learning model for training of the machine learning model; forming a validation training set of images of the other insurance cards classified as associated or not associated with the respective insurance plan or the respective insurance company; and providing the validation training set of images of the other insurance cards to the machine learning model for training of the machine learning model.

“8. The system of claim 1, wherein the user interface of the mobile device of the user displays a comparison between the first insurance plan of the user and the at least one alternative insurance plan for the user.

“9. The system of claim 1, wherein the user interface of the mobile device of the user displays a comparison between a pharmacy of the user and one or more alternative pharmacies for the user, and wherein the comparison comprises one or more of: (i) respective prescription co-payment costs associated with each pharmacy, (ii) a respective savings associated with each of the one or more alternative pharmacies relative to the pharmacy of the user, (iii) a respective proximity of each pharmacy to the user, (iv) a respective network associated with each pharmacy, or (v) respective insurance plan benefits associated with each pharmacy.

“10. The system of claim 1, wherein: the recommendation further comprises: one or more alternative pharmacies for the user to consider as alternatives to a pharmacy of the user, wherein the one or more alternative pharmacies are based on an analysis of one or more of: (i) respective prescription co-payment costs associated with each pharmacy, (ii) a respective savings associated with each of the one or more alternative pharmacies relative to the pharmacy of the user, (iii) a respective proximity of each pharmacy to the user, (iv) a respective network associated with each pharmacy, or (v) respective insurance plan benefits associated with each pharmacy; and extracting the one or more feature values from the at least one image of the insurance card of the user comprises: using (i) one or more text recognition algorithms, (ii) one or more object recognition algorithms, (iii) one or more feature recognition algorithms, or (iv) some combination thereof.

“11. A method being implemented via execution of computing instructions configured to run on one or more processors and stored at one or more non-transitory computer-readable media, the method comprising: extracting one or more feature values from at least one image of an insurance card of a user, wherein the at least one image is scanned from a mobile device of the user; generating a feature vector associated with the at least one image of the insurance card of the user; reducing, using dimensionality reduction, an amount of data in the feature vector to a reduced set of data; determining, by a machine learning model, a first insurance plan of the user based on machine learning model input data comprising the reduced set of data and not comprising the amount of data that is absent from the reduced set of data; identifying at least one alternative insurance plan for the user; and sending instructions to display a recommendation of the at least one alternative insurance plan on a user interface of the mobile device of the user.

“12. The method of claim 11 further comprising: training the machine learning model by: using a set of images of insurance cards as training data for the machine learning model, wherein inputs to the machine learning model comprise historical data for one or more classifications associated with one or more types of insurance cards during a historical time period, wherein outputs of the machine learning model comprise at least one insurance plan matching at least one classification of the one or more classifications associated with the one or more types of insurance cards, and wherein each classification of the one or more classifications is associated with a respective insurance plan.”

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

URL and more information on this patent application, see: Medous, Frederic; Smith, Walter R. Using Machine Learning To Classify Insurance Card Information. U.S. Patent Application Number 20230177617, filed January 30, 2023 and posted June 8, 2023. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20230177617)&db=US-PGPUB&type=ids

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