Survey and suggestion system

Patent metadata

Field	Value
Publication number	US-12567482-B2
Application number	US-202318204798-A
Country	US
Kind code	B2
Filing date	Jun 1, 2023
Priority date	Jun 2, 2022
Publication date	Mar 3, 2026
Grant date	Mar 3, 2026

Abstract

Apparatuses, systems, and methods relate to technology to receive a user request from a user device, identify that a plurality of questions is available for presentation on the user device based on the user request and select a first question from the plurality of questions. The technology transmits the first question to the user device, receives a first user reply from the user device, where the first user reply is an answer to the first question and selects a second question from the plurality of questions based on the first user reply. The technology transmits the second question to the user device, receives a second user reply from the user device, where the second user reply is an answer to the second question, determines a user suggestion based on the first and second user replies, and transmits the user suggestion to the user device.

First claim

We claim:

1 . A computing system comprising:
a processor; and a memory having a set of instructions, which when executed by the processor, cause the computing system to: train a machine learning question and recommendation model (MLQRM) to map recommendations into a vector space as recommendation vectors, wherein to train the MLQRM, the set of instructions when executed, cause the computing system to train the MLQRM based on user responses to questions by patients, recommendations to the patients, and engagement data identifying engagements of the patients with the recommendations, wherein the recommendations mitigate health conditions; receive a user request from a user device; identify that a plurality of questions is available for presentation on the user device based on the user request; select a first question from the plurality of questions; transmit the first question to the user device; receive a first user reply from the user device, wherein the first user reply is an answer to the first question; generate, with a generative artificial intelligence model of the MLQRM, a second question based on the first user reply and the first question; transmit the second question to the user device; receive a second user reply from the user device, wherein the second user reply is an answer to the second question; generate, with the MLQRM, a user vector based on the first and second user replies; determine, with the MLQRM, a user suggestion from the plurality of recommendations based on a similarity between the recommendation vectors and the user vector; and transmit the user suggestion to the user device.

2 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
transmit the first question, the second question, and the user suggestion to the user device via an application programming interface.

3 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
receive the first user reply and the second user reply via an application programming interface.

4 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
execute an authentication process on a current user associated with the user device to authenticate the current user; and determine that communication with the current user is to be executed in response to the current user being authenticated.

5 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
encrypt the first question, the second question, and the user suggestion prior to transmission to the user device.

6 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
decrypt data from the user device to extract the first user reply and the second user reply.

7 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
gather the user responses, and the engagement data; and; translate, with the MLQRM, the user responses into the vector space to generate recommendation vectors.

8 . The computing system of claim 7 , wherein to determine the user suggestion, the MLQRM is configured to:
calculate one or more differences between the recommendation vectors and the user vectors; and select the user suggestion based on the one or more differences.

9 . The computing system of claim 8 , wherein to determine the user suggestion, the MLQRM is configured to:
identify a lowest difference from the one or more differences, wherein the user suggestion is associated with the lowest difference.

10 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
retrieve user data, wherein the user data includes implicit feedback associated with a current user and explicit feedback associated with the current user, wherein the current user is associated with the user device, and further wherein the MLQRM determines the user suggestion further based on the implicit feedback and the explicit feedback.

11 . The computing system of claim 1 , wherein the generative artificial intelligence model generates the second question, with a variational autoencoder.

12 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
determine that the second user reply is to be converted into an observation, wherein the observation is a qualitative representation of the second user reply; initiate an asynchronous state machine based on the second user reply being determined to be converted into the observation; convert the second user reply into the observation with the asynchronous state machine; and cause a third question from the plurality of questions to be presented on the user device as the asynchronous state machine is executed.

13 . The computing system of claim 1 , wherein the user suggestion includes one or more of a first suggested service, a first application, or a first website.

14 . The computing system of claim 13 , wherein the instructions of the memory, when executed, cause the computing system to:
determine that one or more of a second service, a second application, or a second website is to be bypassed for the user suggestion.

15 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
access one or more of rules, a linear ordering, or a state machine to select the first question.

16 . The computing system of claim 1 , wherein the instructions of the memory, when executed, cause the computing system to:
identify user information associated with a current user, wherein the current user is associated with the user device, wherein the user information includes one or more of identifying information of the current user or user measurements of the current user; and determine whether to select a third question of the plurality of questions for transmission to the user device based on whether a third answer to the third question is contained within the user information.

17 . The computing system of claim 1 , wherein the first question, the second question, the first user reply, and the second user reply comply with a Fast Healthcare Interoperability Resources standard.

18 . The computing system of claim 1 , wherein to determine the user suggestion, the set of instructions, which when executed by the processor, cause the computing system to identify the user suggestion with the generative machine learning model.

19 . At least one non-transitory computer readable storage medium comprising a set of instructions, which when executed by a computing system, cause the computing system to:
train a machine learning question and recommendation model (MLQRM) to map recommendations into a vector space as recommendation vectors, wherein to train the MLQRM, the set of instructions when executed, cause the computing system to train the MLQRM based on user responses to questions by patients, recommendations to the patients, and engagement data identifying engagements of the patients with the recommendations, wherein the recommendations mitigate health conditions; receive a user request from a user device; identify that a plurality of questions is available for presentation on the user device based on the user request;

Assignees

• Evernorth Strategic Dev Inc

Inventors

• Mitjans Milton D
• Herzig Kevin

Classifications

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/088

  Non-supervised learning, e.g. competitive learning · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/047

  Probabilistic or stochastic networks · CPC title