Granted Patent B2

US 12,725,086 · App. 19/057,579 · Granted Sep 1, 2026

Inventor: Steven David Alperin (New York, NY)

Assignee: SurvivorNet, Inc.

G06N20/00G06F16/24578G06F16/248

Abstract

Claims (73)

1 . An apparatus for generating data intelligence, the apparatus comprising:

a display device, wherein the display device comprises a remote device communicatively connected to the apparatus;

a processor; and

a memory communicatively connected to the processor, wherein the memory contains instructions configuring the processor to:

receive from a data repository a plurality of digital records, wherein each digital record of the plurality of digital records comprises a plurality of reference attributes;

receive query data from an entity, using the remote device, wherein the query data comprise a plurality of query attributes;

identify one or more relevant digital records by matching one or more query attributes of the plurality of query attributes with one or more reference attributes of the plurality of reference attributes;

quantify relevant features of the one or more relevant digital records using an attention mechanism, wherein the attention mechanism is configured to detect where the relevant features are concentrated;

generate one or more context vectors based on the concentrated relevant features;

generate, using an output generation machine-learning model trained on output generation training data, one or more output data structures as a function of the one or more context vectors, the relevant features, and the one or more relevant digital records;

perform, using at least one of the processor and the output generation machine-learning model, a data augmentation process for generating and adding synthetic data to the output generation training data;

specify, using the one or more output data structures, a treatment plan by populating a hypothetical digital record associated with a hypothetical entity profile with at least one parameter associated with the one or more relevant digital records,

wherein the one or more output data structures are configured to comprise the generated synthetic data;

generate an intelligence machine-learning model, wherein generating the intelligence machine-learning model comprises iteratively training the intelligence machine-learning model using intelligence training data, wherein the intelligence training data correlating one or more output data structures generated using the output generation machine-learning model to intelligence metric inputs;

calculate one or more intelligence metrics, using the trained intelligence machine-learning model, as a function of each output data structure of the one or more output data structures, wherein at least an intelligence metric of the one or more intelligence metrics comprises an estimated likelihood of positive outcome; and wherein the one or more intelligence metrics comprises a frequency metric configured to reflect a technical difficulty level of a process;

select at least a recommended output data structure as a function of the one or more intelligence metrics; and

display, using a user interface, the selected at least a recommended output data.

2 . The apparatus of claim 1 , wherein the processor is further configured to display, using a graphical user interface, the at least a recommended output data structure.

3 . The apparatus of claim 1 , wherein selecting the at least a recommended output data structure comprises:

calculating an aggregate intelligence metric for each output data structure of the one or more output data structures, as a function of the one or more intelligence metrics;

ranking the one or more output data structures as a function of the aggregate intelligence metric; and

selecting the at least a recommended output data structure as a function of the rank.

4 . The apparatus of claim 1 , wherein:

the plurality of digital records comprises a plurality of electronic health records (EHR); and

the plurality of query attributes comprises at least a diagnostic feature pertaining to the entity.

5 . The apparatus of claim 1 , wherein:

the plurality of query attributes comprises at least a resource attribute and at least a positional attribute; and

selecting the at least a recommended output data structure comprises identifying the at least a recommended output data structure as a function of the at least a resource attribute and the at least a positional attribute.

6 . The apparatus of claim 1 , wherein:

receiving the plurality of digital records comprises grouping the plurality of digital records into a plurality of cohorts, wherein each cohort of the plurality of cohorts shares one or more reference attributes; and

calculating the one or more intelligence metrics comprises:

labeling each output data structure of the one or more output data structures as a function of the plurality of cohorts; and

calculating the one or more intelligence metrics as a function of the label.

7 . The apparatus of claim 1 , wherein generating the one or more output data structures comprises:

receiving output generation training data comprising a plurality of exemplary output data structures as outputs correlated with a plurality of exemplary digital records as inputs;

iteratively training the output generation machine-learning model using the output generation training data; and

generating the one or more output data structures using the output generation machine-learning model.

8 . The apparatus of claim 1 , wherein at least an intelligence metric of the one or more intelligence metrics comprises at least a risk metric.

9 . A method for generating data intelligence, the method comprising: receiving, by a processor from a data repository, a plurality of digital records, wherein

each digital record of the plurality of digital records comprises a plurality of reference attributes;

receiving, by the processor from an entity, query data, wherein the query data comprise a plurality of query attributes;

identifying, by the processor, one or more relevant digital records by matching one or more query attributes of the plurality of query attributes with one or more reference attributes of the plurality of reference attributes;

quantifying, by the processor, relevant features of the one or more relevant digital records using an attention mechanism, wherein the attention mechanism is configured to detect where the relevant features are concentrated;

generating, by the processor, one or more context vectors based on the concentrated relevant features;

generating, by the processor using an output generation machine-learning model trained on output generation training data, one or more output data structures as a function of the one or more context vectors, the relevant features, and the one or more relevant digital records;

performing, using at least one of the processor and the output generation machine-learning model, a data augmentation process for generating and adding synthetic data to the output generation training data;

specify, using the one or more output data structures, a treatment plan by populating a hypothetical digital record associated with a hypothetical entity profile with at least one parameter associated with the one or more relevant digital records,

wherein the one or more output data structures are configured to comprise the generated synthetic data;

generating, by the at least a processor, an intelligence machine-learning model, wherein generating the intelligence machine-learning model comprises iteratively training the intelligence machine-learning model using intelligence training data, wherein the intelligence training data correlating one or more output data structures generated using the output generation machine-learning model to intelligence metric inputs;

calculating, by the processor, one or more intelligence metrics using the trained intelligence machine-learning model, as a function of each output data structure of the one or more output data structures, wherein at least an intelligence metric of the one or more intelligence metrics comprises an estimated likelihood of positive outcome; and wherein the one or more intelligence metrics comprises a frequency metric configured to reflect a technical difficulty level of a process;

selecting, by the processor, at least a recommended output data structure as a function of the one or more intelligence metrics; and

displaying, by a user interface, the selected at least a recommended output data.

10 . The method of claim 9 , further comprising displaying, by the processor using a graphical user interface, the at least a recommended output data structure.

11 . The method of claim 9 , wherein selecting the at least a recommended output data structure comprises:

calculating an aggregate intelligence metric for each output data structure of the one or more output data structures, as a function of the one or more intelligence metrics;

ranking the one or more output data structures as a function of the aggregate intelligence metric; and

selecting the at least a recommended output data structure as a function of the rank.

12 . The method of claim 9 , wherein:

the plurality of digital records comprises a plurality of electronic health records (EHR); and

the plurality of query attributes comprises at least a diagnostic feature pertaining to the entity.

13 . The method of claim 9 , wherein:

the plurality of query attributes comprises at least a resource attribute and at least a positional attribute; and

selecting the at least a recommended output data structure comprises identifying the at least a recommended output data structure as a function of the at least a resource attribute and the at least a positional attribute.

14 . The method of claim 9 , wherein:

receiving the plurality of digital records comprises grouping the plurality of digital records into a plurality of cohorts, wherein each cohort of the plurality of cohorts shares one or more reference attributes; and

calculating the one or more intelligence metrics comprises:

labeling each output data structure of the one or more output data structures as a function of the plurality of cohorts; and

calculating the one or more intelligence metrics as a function of the label.

15 . The method of claim 9 , wherein generating the one or more output data structures comprises:

receiving output generation training data comprising a plurality of exemplary output data structures as outputs correlated with a plurality of exemplary digital records as inputs;

iteratively training the output generation machine-learning model using the output generation training data; and

generating the one or more output data structures using the output generation machine-learning model.

16 . The method of claim 9 , wherein at least an intelligence metric of the one or more intelligence metrics comprises at least a risk metric.

Assignments (1)

Continuity (1)

References Cited (17)