IP Library › Granted Patent US 12664445

Granted Patent B2

US 12664445 · App. 16/538,129 · Granted Jun 23, 2026

Predictive data analysis in conceptually hierarchical domains

Inventors: Jason Robinson (La Jolla, CA); Mariya Wahlstrom (Escondido, CA); Ellyn Oliver (San Diego, CA); Michael Pedersen (San Diego, CA); Mark Messer (Escondido, CA); Brian C. Potter (Carlsbad, CA); Kelly Canter (Roscommon, MI); Mark L. Morsch (San Diego, CA)

Assignee: Optum, Inc.

G06N5/04 G06N5/02

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12664445
App. No.: 16/538,129
Granted: Jun 23, 2026
Kind: B2

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

There is a need for solutions for more effective and efficient predictive data analysis systems in conceptually hierarchical domains. This need can be addressed, for example, by a system configured to obtain one or more initial raw inputs; determine a partial prediction for the one or more initial raw inputs, wherein the partial prediction is associated with an initial encoding hierarchy and the initial encoding hierarchy is associated with a plurality of prediction nodes; determine, based on the partial prediction and the initial encoding hierarchy, one or more partial prediction information deficiencies for partial prediction; obtain one or more supplemental raw inputs based on the one or more partial prediction information deficiencies; and generate a conceptually hierarchical prediction based on the one or more supplemental raw inputs and the partial prediction.

Claims (112)

1 . A computer-implemented method comprising:

receiving, by one or more processors, initial raw data input originating from a computing device;

generating, by the one or more processors and based at least in part on natural language processing performed on the initial raw data input, an inferred prediction value and a prediction score for each of a plurality of prediction nodes stored within a memory storage area as an encoding hierarchy that defines hierarchical relationships among the plurality of prediction nodes;

generating, by the one or more processors and in the memory storage area, a partial code prediction data object comprising one or more values and a first null value of a partial medical code, wherein the partial code prediction data object comprises data generated based at least in part on output generated by a subset of the plurality of prediction nodes having respective prediction scores satisfying a prediction score criterion within the encoding hierarchy;

wherein the subset of the plurality of prediction nodes are linked with one another in the memory storage area along a partial prediction traversal path determined at least in part by:

automatically selecting a plurality of second prediction nodes using links defining a hierarchical dependency from a first prediction node within the memory storage area when the first null value is generated, using the first prediction node of the subset of the plurality of prediction nodes, based at least in part on output of the natural language processing performed on the initial raw data input;

determining, by the one or more processors in real-time prior to generating a final code prediction data object and based at least in part on the encoding hierarchy, a partial code prediction data deficiency for the partial code prediction data object based at least in part on the first null value of the partial code prediction data object;

generating, by the one or more processors, a data request in real-time upon determining the partial code prediction data deficiency, wherein the data request includes one or more parameters determined from the partial code prediction data deficiency;

transmitting, by the one or more processors, the data request to the computing device, causing the computing device to generate at least one of an audio output or a visual output to prompt a user associated with the computing device for supplemental raw data input based at least in part on the one or more parameters;

receiving, by the one or more processors and from the computing device, the supplemental raw data input based at least in part on the data request; and

generating, by the one or more processors, the final code prediction data object, wherein the final code prediction data object is a complete medical code that comprises the one or more values and a final value replacing the first null value of the partial code prediction data object based at least in part on output of natural language processing performed on the supplemental raw data input and the partial code prediction data object.

2 . The computer-implemented method of claim 1 , wherein the hierarchical relationships comprise one or more predictive dependence relationships among the plurality of prediction nodes.

3 . The computer-implemented method of claim 2 , wherein the one or more predictive dependence relationships comprise at least one predictive dependence relationship having a necessity predictive semantic and at least one predictive dependence relationship having a disjunctive predictive semantic.

4 . The computer-implemented method of claim 1 , further comprising:

determining the encoding hierarchy based at least in part on the initial raw data input.

5 . The computer-implemented method of claim 4 , wherein determining the encoding hierarchy based at least in part on the initial raw data input comprises:

receiving one or more encoding hierarchies from the memory storage area;

for each encoding hierarchy of the one or more encoding hierarchies, determining a hierarchy prediction score based at least in part on the initial raw data input; and

selecting a first encoding hierarchy of the one or more encoding hierarchies having a highest prediction score as the encoding hierarchy.

6 . The computer-implemented method of claim 1 , wherein the initial raw data input comprises one or more medical provider inputs.

7 . The computer-implemented method of claim 1 , wherein the initial raw data input comprises one or more natural language inputs.

8 . The computer-implemented method of claim 1 , wherein the partial prediction traversal path is a portion of a selected prediction traversal path of one or more prediction traversal paths of the encoding hierarchy, and wherein generating the partial code prediction data object comprises:

determining an ontology data object based at least in part on the initial raw data input; and

determining the selected prediction traversal path of the one or more prediction traversal paths for the ontology data object.

9 . The computer-implemented method of claim 8 , wherein determining the ontology data object comprises:

receiving, as a part of the initial raw data input, one or more inferred prediction values;

determining, for each inferred prediction value of the one or more inferred prediction values, a value prediction score;

determining one or more reliable prediction values of the one or more inferred prediction values whose value prediction score exceeds a value prediction score threshold; and

determining the ontology data object based at least in part on the one or more reliable prediction values.

10 . The computer-implemented method of claim 1 , further comprising:

determining the first prediction node, wherein the first prediction node has a positive prediction value;

determining the plurality of second prediction nodes, wherein the plurality of second prediction nodes are hierarchically dependent from the first prediction node;

determining a confidence score for each of a plurality of values generated corresponding to the first prediction node; and

selecting a first value of the plurality of values generated corresponding to the first prediction node as having a highest confidence score.

11 . The computer-implemented method of claim 1 , wherein causing the computing device to generate at least one of an audio output or a visual output to prompt a user associated with the computing device for supplemental raw data input based at least in part on the one or more parameters comprises:

causing the computing device to generate an audio output using the one or more parameters, one or more semantic rules, and a voice synthesis routine.

12 . The computer-implemented method of claim 1 , further comprising:

determining a hierarchical closure indication for the final code prediction data object based at least in part on the final code prediction data object and the encoding hierarchy;

determining whether the hierarchical closure indication indicates that the final code prediction data object has hierarchical closure; and

based at least in part on the determining whether the hierarchical closure indication indicates that the final code prediction data object has hierarchical closure:

in response to determining that the hierarchical closure indication indicates that the final code prediction data object has hierarchical closure, transmitting the final code prediction data object to a recipient computing entity; or

in response to determining that the hierarchical closure indication indicates that the final code prediction data object does not have hierarchical closure,

determining one or more final prediction data deficiencies based at least in part on the final code prediction data object and the encoding hierarchy,

generating a supplemental audio output using one or more supplemental data parameters, one or more semantic rules, and a voice synthesis routine, and

transmitting the supplemental audio output to a data provider computing entity.

13 . The computer-implemented method of claim 1 , wherein the encoding hierarchy is a medical code hierarchy and wherein the partial code prediction data object reflects information deficiencies in generating a medical code.

14 . The computer-implemented method of claim 1 , wherein the one or more parameters determined from the partial code prediction data deficiency identifies supplemental information needed to generate a complete medical code as the final code prediction data object.

15 . An apparatus comprising at least one processor and at least one non-transitory memory comprising program code, wherein the at least one non-transitory memory and the program code are configured to, with the at least one processor, cause the apparatus to:

receive an initial raw data input originating from a computing device;

generate, based at least in part on natural language processing performed on the initial raw data input, an inferred prediction value and a prediction score for each of a plurality of prediction nodes stored within the at least one non-transitory memory as an encoding hierarchy that defines hierarchical relationships among the plurality of prediction nodes;

generate, in the at least one non-transitory memory, a partial code prediction data object comprising one or more values and a first null value of a partial medical code, wherein the partial code prediction data object comprises data generated based at least in part on output generated by a subset of the plurality of prediction nodes having respective prediction scores satisfying a prediction score criterion within the encoding hierarchy;

wherein the subset of the plurality of prediction nodes are linked with one another in the at least one non-transitory memory along a partial prediction traversal path and wherein to determine the partial prediction traversal path, the at least one non-transitory memory and the program code are configured to, with the at least one processor, cause the apparatus to:

automatically select a plurality of second prediction nodes using links defining a hierarchical dependency from a first prediction node within the at least one non-transitory memory when the first null value is generated, using the first prediction node of the subset of the plurality of prediction nodes, based at least in part on output of the natural language processing performed on the initial raw data input;

determine, in real-time before the apparatus generates a final code prediction data object and based at least in part on the encoding hierarchy, a partial code prediction data deficiency for the partial code prediction data object based at least in part on the first null value of the partial code prediction data object;

generate a data request in real-time when the apparatus determines the partial code prediction data deficiency, wherein the data request includes one or more parameters determined from the partial code prediction data deficiency;

transmit the data request to the computing device to cause the computing device to generate at least one of an audio output or a visual output to prompt a user associated with the computing device for supplemental raw data input based at least in part on the one or more parameters;

receive the supplemental raw data input from the computing device based at least in part on the data request; and

generate the final code prediction data object, wherein the final code prediction data object is a complete medical code that comprises the one or more values and a final value replacing the first null value of the partial code prediction data object based at least in part on output of natural language processing performed on the supplemental raw data input and the partial code prediction data object.

16 . The apparatus of claim 15 , wherein the partial prediction traversal path is a portion of a selected prediction traversal path of one or more prediction traversal paths of the encoding hierarchy, and wherein to generate the partial code prediction data object, the apparatus is further configured to:

determine an ontology data object based at least in part on the initial raw data input; and

determine the selected prediction traversal path of the one or more prediction traversal paths for the ontology data object.

17 . The apparatus of claim 16 , wherein to determine the ontology data object, the apparatus is further configured to:

receive, as a part of the initial raw data input, one or more inferred prediction values;

determine, for each inferred prediction value of the one or more inferred prediction values, a value prediction score;

determine one or more reliable prediction values of the one or more inferred prediction values whose value prediction score exceeds a value prediction score threshold; and

determine the ontology data object based at least in part on the one or more reliable prediction values.

18 . The apparatus of claim 15 , wherein the apparatus is further configured to:

determine the first prediction node, wherein the first prediction node has a positive prediction value,

determine the plurality of second prediction nodes, wherein the plurality of second prediction nodes are hierarchically dependent from the first prediction node, determine a confidence score for each of a plurality of values generated corresponding to the first prediction node; and

select a first value of the plurality of values generated corresponding to the first prediction node as having a highest confidence.

19 . The apparatus of claim 15 , wherein the at least one non-transitory memory and the program code are further configured to, with the at least one processor, cause the apparatus to:

determine a hierarchical closure indication for the final code prediction data object based at least in part on the final code prediction data object and the encoding hierarchy;

determine whether the hierarchical closure indication indicates that the final code prediction data object has hierarchical closure; and

based at least in part on whether the hierarchical closure indication indicates that the final code prediction data object has hierarchical closure:

when the apparatus determines that the hierarchical closure indication indicates that the final code prediction data object has hierarchical closure, transmit the final code prediction data object to a recipient computing entity; or

when the apparatus determines that the hierarchical closure indication indicates that the final code prediction data object does not have hierarchical closure,

determine one or more final prediction data deficiencies based at least in part on the final code prediction data object and the encoding hierarchy,

generate a supplemental data voice prompt using one or more supplemental data parameters, one or more semantic rules, and a voice synthesis routine, and

transmit the supplemental data voice prompt to a data provider computing entity.

20 . A non-transitory computer storage medium comprising instructions configured to cause one or more processors to at least:

receive an initial raw data input originating from a computing device;

generate, based at least in part on natural language processing performed on the initial raw data input, an inferred prediction value and a prediction score for each of a plurality of prediction nodes stored within a memory as an encoding hierarchy that defines hierarchical relationships among the plurality of prediction nodes;

generate, in the memory, a partial code prediction data object comprising one or more values and a first null value of a partial medical code, wherein the partial code prediction data object comprises data generated based at least in part on output generated by a subset of the plurality of prediction nodes having respective prediction scores satisfying a prediction score criterion within the encoding hierarchy;

wherein the subset of the plurality of prediction nodes are linked with one another in the memory along a partial prediction traversal path and wherein to determine the partial prediction traversal path, the instructions are further configured to cause one or more processors to at least:

automatically select a plurality of second prediction nodes using links defining a hierarchical dependency from a first prediction node within the memory when the first null value is generated, using the first prediction node of the subset of the plurality of prediction nodes, based at least in part on output of the natural language processing performed on the initial raw data input;

determine, in real-time before the one or more processors generate a final code prediction data object and based at least in part on the encoding hierarchy, a partial code prediction data deficiency for the partial code prediction data object based at least in part on the first null value of the partial code prediction data object;

generate a data request in real-time when the one or more processors determine the partial code prediction data deficiency, wherein the data request includes one or more parameters determined from the partial code prediction data deficiency;

receive the supplemental raw data input from the computing device based at least in part on the data request; and

generate the final code prediction data object, wherein the final code prediction data object is a complete medical code that comprises the one or more values and a final value replacing the first null value of the partial code prediction data object based at least in part on natural language processing performed on the supplemental raw data input and the partial code prediction data object.

21 . The non-transitory computer storage medium of claim 20 , wherein the partial prediction traversal path is a portion of a selected prediction traversal path of one or more prediction traversal paths of the encoding hierarchy, and wherein to generate the partial code prediction data object, the instructions are further configured to cause one or more processors to at least:

determine an ontology data object based at least in part on the initial raw data input; and

determine the selected prediction traversal path of the one or more prediction traversal paths for the ontology data object.

22 . The non-transitory computer storage medium of claim 21 , wherein to determine the ontology data object the instructions are further configured to cause one or more processors to at least:

process the initial raw data input to receive one or more inferred prediction values;

determine, for each inferred prediction value of the one or more inferred prediction values, a value prediction score;

determine one or more reliable prediction values of the one or more inferred prediction values whose value prediction score exceeds a value prediction score threshold; and

determine the ontology data object based at least in part on the one or more reliable prediction values.

23 . The non-transitory computer storage medium of claim 20 , wherein the instructions are further configured to cause one or more processors to at least:

determine the first prediction node, wherein the first prediction node has a positive prediction value,

determine the plurality of second prediction nodes, wherein the plurality of second prediction nodes are hierarchically dependent from the first prediction node,

determine a confidence score for each of a plurality of values generated corresponding to the first prediction node; and

select a first value of the plurality of values generated corresponding to the first prediction node as having a highest confidence score.

24 . The non-transitory computer storage medium of claim 20 , wherein the instructions are further configured to cause one or more processors to at least: