# US 2026/0073428 A1 · Systems and Methods For Preparation And Retrieval Of User-Level Treatment Recommendations From A Machine Learning Model

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0073428 A1
- Document ID: US20260073428A1
- Status: Published application
- Kind: A1
- Application number: 19/070,383
- Filed: 2025-03-04
- Published: 2026-03-12
- Earliest claimed: 2024-09-06
- Assignees: Auxia, Inc.
- Applicants: Auxia, Inc.
- Inventors: Ravi Desu; Sumeet Kumar; Asim Krishna Prasad; Yichen Zhao
- Web page: https://patentgrep.com/patent/US20260073428A1

## CPC

- G06N 20/00: Machine learning
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06Q 30/0255: {based on user history}
- G06Q 30/0269: {based on user profile or attribute}
- G06Q 30/0271: {Personalized advertisement}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Disclosed herein is a computerized method including operations of receiving a request for a treatment recommendation, where the request includes a user ID and where a treatment is a notification, alert, or message to be provided by a network device. The operations further include retrieving a set of user data from an aggregated user data table according to the user ID and treatment data comprised of data related to a set of predefined treatments from a treatment datastore, determine a score for each of the set of candidate treatments through processing of the set of user data, a set of candidate treatments, and a configured objective as input by a machine learning model, generating a ranked list of treatments based on the scoring provided by the machine learning model, and providing the ranked list of the set of candidate treatments and the set of candidate treatments to the network device.

## Claims

### Claim 1

A computerized method comprising:
receiving a request for a treatment recommendation, wherein the request includes a user identifier (ID), and wherein a treatment is a notification, alert, or message to be displayed by a network device;
retrieving (i) a set of user data from an aggregated user data table according to the user ID, and (ii) treatment data comprised of data related to a set of predefined treatments from a treatment datastore;
deploying a machine learning model configured to take the set of user data, a set of candidate treatments, and a configured objective as input and provide a scoring for each of the set of candidate treatments, wherein the set of candidate treatments is selected from the set of predefined treatments according to application of a set of eligibility rules, and wherein the machine learning model is trained and configured to score each of the set of candidate treatments according to the configured objective;
generating a ranked list of the set of candidate treatments based on the scoring provided by the machine learning model; and
providing the ranked list of the set of candidate treatments and the set of candidate treatments to the network device.

### Claim 2 (depends on 1)

The computerized method of claim 1, wherein the request is received from a software application running on the network device.

### Claim 3 (depends on 1)

The computerized method of claim 1, wherein selecting the set of candidate treatments includes applying the set of eligibility rules to the set of user data such that the set of candidate treatments is a subset of the set of predefined treatments.

### Claim 4 (depends on 1)

The computerized method of claim 1, wherein the ranked list of the set of candidate treatments provided to the network device is stored in a treatment history datastore.

### Claim 5 (depends on 1)

The computerized method of claim 1, wherein generating the ranked list of the set of candidate treatments includes applying a set of customer-specific ranking logic that alter a ranking of one or more of the set of candidate treatments.

### Claim 6 (depends on 1)

The computerized method of claim 1, wherein generating the ranked list of the set of candidate treatments includes applying a set of mutual exclusion rules that prevents multiple treatments included within a predefined grouping from being provided to the network device.

### Claim 7 (depends on 1)

The computerized method of claim 1, wherein the scoring provided by the machine learning model is indicative of a likelihood that each of the set of candidate treatments will prompt a user of the network device to take a desired action.

### Claim 8 (depends on 1)

The computerized method of claim 1, wherein the set of user data is aggregated and transformed into a set of vector embeddings by machine learning models.

### Claim 9 (depends on 1)

The computerized method of claim 1, wherein the treatment data is aggregated and transformed into a set of vector embeddings by machine learning models.

### Claim 10 (depends on 1)

The computerized method of claim 1, wherein an assignment of a user to a control group or a treatment group is made after the treatment scoring and selection process has completed for that user.

### Claim 11 (depends on 1)

The computerized method of claim 1, wherein a state of the set of user data at a time that one or more treatments is scored, selected and delivered to a user of the network device is recorded and stored.

### Claim 12 (depends on 1)

The computerized method of claim 1, wherein the treatment includes a graphical image.

### Claim 13

A computing device, comprising:
a processor; and
a non-transitory computer-readable medium having stored thereon instructions that, when executed by the processor, cause the processor to perform operations including:
receiving a request for a treatment recommendation, wherein the request includes a user identifier (ID), and wherein a treatment is a notification, alert, or message to be provided by a network device,
retrieving (i) a set of user data from an aggregated user data table according to the user ID, and (ii) treatment data comprised of data related to a set of predefined treatments from a treatment datastore,
deploying a machine learning model configured to take the set of user data, a set of candidate treatments, and a configured objective as input and provide a scoring for each of the set of candidate treatments, wherein the set of candidate treatments is selected from the set of predefined treatments according to application of a set of eligibility rules, and wherein the machine learning model is trained and configured to score each of the set of candidate treatments according to the configured objective,
generating a ranked list of the set of candidate treatments based on the scoring provided by the machine learning model, and
providing the ranked list of the set of candidate treatments and the set of candidate treatments to the network device.

### Claim 14 (depends on 13)

The computing device of claim 13, wherein selecting the set of candidate treatments includes applying the set of eligibility rules to the set of user data such that the set of candidate treatments is a subset of the set of predefined treatments.

### Claim 15 (depends on 13)

The computing device of claim 13, wherein the scoring provided by the machine learning model is indicative of a likelihood that each of the set of candidate treatments will prompt a user of the network device to take a desired action.

### Claim 16 (depends on 13)

The computing device of claim 13, wherein the set of user data is aggregated and transformed into a set of vector embeddings by machine learning models, and wherein the treatment data is aggregated and transformed into a set of vector embeddings by machine learning models.

### Claim 17

A non-transitory computer-readable medium having stored thereon instructions that, when executed by one or more processors, cause the one or more processor to perform operations including:
receiving a request for a treatment recommendation, wherein the request includes a user identifier (ID), and wherein a treatment is a notification, alert, or message to be provided by a network device;
retrieving (i) a set of user data from an aggregated user data table according to the user ID, and (ii) treatment data comprised of data related to a set of predefined treatments from a treatment datastore;
deploying a machine learning model configured to take the set of user data, a set of candidate treatments, and a configured objective as input and provide a scoring for each of the set of candidate treatments, wherein the set of candidate treatments is selected from the set of predefined treatments according to application of a set of eligibility rules, and wherein the machine learning model is trained and configured to score each of the set of candidate treatments according to the configured objective;
generating a ranked list of the set of candidate treatments based on the scoring provided by the machine learning model; and
providing the ranked list of the set of candidate treatments and the set of candidate treatments to the network device.

### Claim 18 (depends on 17)

The non-transitory computer-readable medium of claim 17, wherein selecting the set of candidate treatments includes applying the set of eligibility rules to the set of user data such that the set of candidate treatments is a subset of the set of predefined treatments.

### Claim 19 (depends on 17)

The non-transitory computer-readable medium of claim 17, wherein the scoring provided by the machine learning model is indicative of a likelihood that each of the set of candidate treatments will prompt a user of the network device to take a desired action.

### Claim 20 (depends on 17)

The non-transitory computer-readable medium of claim 17, wherein the set of user data is aggregated and transformed into a set of vector embeddings by machine learning models, and wherein the treatment data is aggregated and transformed into a set of vector embeddings by machine learning models.
