Granted Patent B2

US 12,531,157 · App. 18/602,222 · Granted Jan 20, 2026

Inventors: Bo Wen (Chappaqua, NY); Chen Wang (Chappaqua, NY)

Assignee: International Business Machines Corporation

G16H50/20G06F40/20G16H10/20

Abstract

Claims (72)

1 . A method of processing requests to perform projects comprising:

processing a request in a natural language to perform a project via a hierarchy of machine learning agents of at least one processor, wherein one or more machine learning agents of a management layer of the hierarchy determine and assign tasks for the project to one or more machine learning agents of an operation layer of the hierarchy based on the request;

performing the assigned tasks by the one or more machine learning agents of the operation layer to perform the project;

monitoring operation of the machine learning agents of the operation layer by one or more machine learning agents of an administration layer of the hierarchy; and

adjusting behavior of the machine learning agents of the operation layer by the one or more machine learning agents of the administration layer to reduce one or more from a group of non-compliance incidents and training iterations, wherein the adjusting the behavior of the machine learning agents comprises shortening a number of training iterations and adding a plurality of longest training cases to an evaluation dataset used by the monitoring operation.

2 . The method of claim 1 , further comprising:

monitoring operation of the machine learning agents of the operation layer by one or more machine learning agents of an administration layer of the hierarchy; and

adjusting behavior of the machine learning agents of the operation layer by the one or more machine learning agents of the administration layer in response to one or more from a group of deficient performance and non-compliance with guidelines.

3 . The method of claim 2 ,

wherein the machine learning agents of the administration layer, the management layer, and the operation layer include large language models and the monitoring is based on natural language reports.

4 . The method of claim 3 ,

wherein adjusting behavior of the machine learning agents of the operation layer comprises:

adjusting prompts for the large language models of the operation layer.

5 . The method of claim 3 ,

wherein the machine learning agents of the administration layer include large-sized large language models, the machine learning agents of the management layer include medium-sized large language models, and the machine learning agents of the operation layer include small-sized large language models.

6 . The method of claim 1 , further comprising:

interacting with health care providers in the natural language, via at least one machine learning agent of the management layer, to receive the request; and

interacting with patients in the natural language to obtain information and translating the information obtained from the patients into a medical questionnaire to perform the project via at least one machine learning agent of the operation layer.

7 . The method of claim 6 ,

wherein the patients interact with the at least one machine learning agent of the operation layer via one or more from a group of phone, text messages, smart devices, email, and social media platforms.

8 . The method of claim 1 , wherein the plurality of the longest training cases are identified based on one or more of a number of training iterations, an execution time, size of the assigned tasks, and complexity of the assigned tasks.

9 . A system for processing requests to perform projects comprising:

one or more processors; and

one or more memory devices coupled to the one or more processors, wherein the one or more processors are configured to:

process a request in a natural language to perform a project via a hierarchy of machine learning agents, wherein one or more machine learning agents of a management layer of the hierarchy determine and assign tasks for the project to one or more machine learning agents of an operation layer of the hierarchy based on the request;

perform the assigned tasks by the one or more machine learning agents of the operation layer to perform the project;

monitor operation of the machine learning agents of the operation layer by one or more machine learning agents of an administration layer of the hierarchy; and

adjust behavior of the machine learning agents of the operation layer by the one or more machine learning agents of the administration layer to reduce one or more from a group of non-compliance incidents and training iterations,

wherein the one or more processors, to adjust the behavior of the machine learning agents, are further configured to shorten a number of training iterations and add a plurality of longest training cases to an evaluation dataset used by the monitoring operation.

10 . The system of claim 9 ,

wherein the one or more processors are further configured to:

monitor operation of the machine learning agents of the operation layer by one or more machine learning agents of an administration layer of the hierarchy; and

adjust behavior of the machine learning agents of the operation layer by the one or more machine learning agents of the administration layer in response to one or more from a group of deficient performance and non-compliance with guidelines.

11 . The system of claim 10 ,

wherein the machine learning agents of the administration layer, the management layer, and the operation layer include large language models and the monitoring is based on natural language reports.

12 . The system of claim 11 ,

wherein the one or more processors, to adjust the behavior of the machine learning agents of the operation layer, are further configured to:

adjust prompts for the large language models of the operation layer.

13 . The system of claim 11 ,

wherein the machine learning agents of the administration layer include large-sized large language models, the machine learning agents of the management layer include medium-sized large language models, and the machine learning agents of the operation layer include small-sized large language models.

14 . The system of claim 9 ,

wherein the one or more processors are further configured to:

interact with health care providers in the natural language, via at least one machine learning agent of the management layer, to receive the request; and

interact with patients in the natural language to obtain information and translate the information obtained from the patients into a medical questionnaire to perform the project via at least one machine learning agent of the operation layer.

15 . The system of claim 14 ,

wherein the patients interact with the at least one machine learning agent of the operation layer via one or more from a group of phone, text messages, smart devices, email, and social media platforms.

16 . The system of claim 9 , wherein the plurality of the longest training cases are identified based on one or more of a number of training iterations, an execution time, size of the assigned tasks, and complexity of the assigned tasks.

17 . A non-transitory computer-readable medium storing a set of instructions for wireless communication, the set of instructions comprising:

one or more instructions that, when executed by one or more processors of a device, cause the device to:

process a request in a natural language to perform a project via a hierarchy of machine learning agents, wherein one or more machine learning agents of a management layer of the hierarchy determine and assign tasks for the project to one or more machine learning agents of an operation layer of the hierarchy based on the request;

perform the assigned tasks by the one or more machine learning agents of the operation layer to perform the project;

monitor operation of the machine learning agents of the operation layer by one or more machine learning agents of an administration layer of the hierarchy; and

adjust behavior of the machine learning agents of the operation layer by the one or more machine learning agents of the administration layer to reduce one or more from a group of non-compliance incidents and training iterations,

wherein the one or more instructions, to adjust the behavior of the machine learning agents, cause the device to shorten a number of training iterations and add a plurality of longest training cases to an evaluation dataset used by the monitoring operation.

18 . The non-transitory computer-readable medium of claim 17 ,

wherein the one or more instructions cause the device to:

monitor operation of the machine learning agents of the operation layer by one or more machine learning agents of an administration layer of the hierarchy; and

adjust behavior of the machine learning agents of the operation layer by the one or more machine learning agents of the administration layer in response to one or more from a group of deficient performance and non-compliance with guidelines.

19 . The non-transitory computer-readable medium of claim 18 ,

wherein the machine learning agents of the administration layer, management layer, and the operation layer include large language models and the monitoring is based on natural language reports.

20 . The non-transitory computer-readable medium of claim 19 ,

wherein the one or more instructions, to adjust the behavior of the machine learning agents of the operation layer, cause the device to:

adjust prompts for the large language models of the operation layer.

21 . The non-transitory computer-readable medium of claim 19 ,

wherein the machine learning agents of the administration layer include large-sized large language models, the machine learning agents of the management layer include medium-sized large language models, and the machine learning agents of the operation layer include small-sized large language models.

22 . The non-transitory computer-readable medium of claim 17 ,

wherein the one or more instructions cause the device to:

interact with health care providers in the natural language, via at least one machine learning agent of the management layer, to receive the request; and

interact with patients in the natural language to obtain information and translate the information obtained from the patients into a medical questionnaire to perform the project via at least one machine learning agent of the operation layer.

23 . The non-transitory computer-readable medium of claim 22 ,

wherein the patients interact with the at least one machine learning agent of the operation layer via one or more from a group of phone, text messages, smart devices, email, and social media platforms.

24 . The non-transitory computer-readable medium of claim 17 , wherein the plurality of the longest training cases are identified based on one or more of a number of training iterations, an execution time, size of the assigned tasks, and complexity of the assigned tasks.

Assignments (1)

Continuity (1)

References Cited (26)