Granted Patent B2

US 12,525,354 · App. 17/374,540 · Granted Jan 13, 2026

Inventors: Rama Krishna Singh (Greater Noida, IN); Ravi Pande (Noida, IN); Priyank Jain (Noida, IN)

Assignee: Optum Technology, Inc.

G16H50/20G06F18/23G06F40/20G06N20/00

Abstract

Claims (79)

1 . A computer-implemented method comprising:

identifying, by one or more processors, a plurality of time bound code transition likelihood data objects for an entity cluster that is associated with an identifiable data entity, wherein: (i) a time bound code transition likelihood data object of the plurality of time bound code transition likelihood data objects is associated with a defined time bound of a plurality of defined time bounds, (ii) the time bound code transition likelihood data object for the defined time bound describes, for a code pair comprising a first defined occurrence code of a plurality of defined occurrence codes and a second defined occurrence code of the plurality of defined occurrence codes, an inferred likelihood that the second defined occurrence code occurs within the defined time bound of an assumed occurrence of the first defined occurrence code, and (iii) the time bound code transition likelihood data object is generated for the defined time bound by:

identifying a plurality of per-entity code occurrences for the entity cluster, wherein a per-entity code occurrence of the plurality of per-entity code occurrences describes that a corresponding defined occurrence code occurs in relation to a corresponding identifiable data entity in the entity cluster at a corresponding timestamp, and

generating the time bound code transition likelihood data object based at least in part on the plurality of per-entity code occurrences;

generating, by the one or more processors and based at least in part on the plurality of time bound code transition likelihood data objects and a plurality of individual historic code occurrences for the identifiable data entity, a future occurrence code prediction, for the identifiable data entity, comprising a selected subset of the plurality of defined occurrence codes; and

initiating, by the one or more processors, one or more prediction-based actions based at least in part on the future occurrence code prediction.

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

determining an entity-code occurrence data object for the entity cluster based at least in part on the plurality of per-entity code occurrences; and

generating an augmented entity-code occurrence data object based at least in part on the entity-code occurrence data object.

3 . The computer-implemented method of claim 2 , wherein generating the augmented entity-code occurrence data object comprises:

performing singular value decomposition on the entity-code occurrence data object to generate a plurality of target lower-ranked entity-code occurrence data objects;

determining at least one missing value in the entity-code occurrence data object based at least in part on the plurality of target lower-ranked entity-code occurrence data objects; and

generating the augmented entity-code occurrence data object based at least in part on the at least one missing value.

4 . The computer-implemented method of claim 2 , wherein generating the time bound code transition likelihood data object based at least in part on the plurality of per-entity code occurrences comprises:

determining a plurality of augmented per-entity code occurrences for the entity cluster based at least in part on the plurality of per-entity code occurrences and the augmented entity-code occurrence data object;

determining a time bound code transition document that describes all co-occurrences of code pairs from the plurality of defined occurrence codes associated with the plurality of augmented per-entity code occurrences that occur within the defined time bound; and

generating the time bound code transition likelihood data object based at least in part on the time bound code transition document.

5 . The computer-implemented method of claim 4 , wherein generating the time bound code transition likelihood data object based at least in part on the time bound code transition document comprises:

determining a time bound code transition likelihood object for the code pair based at least in part on a per-document occurrence of the code pair as described by the time bound code transition document for the defined time bound and a cross-document occurrence of the code pair as described by the time bound code transition document; and

generating the time bound code transition likelihood data object based at least in part on the time bound code transition likelihood object.

6 . The computer-implemented method of claim 5 , wherein the time bound code transition likelihood object is a term-frequency-inverse-domain-frequency measure.

7 . The computer-implemented method of claim 1 , wherein generating the selected subset comprises:

generating a first subset of the plurality of defined occurrence codes based at least in part on the plurality of time bound code transition likelihood data objects and the plurality of individual historic code occurrences for the identifiable data entity;

generating a second subset of the plurality of defined occurrence codes based at least in part on a cross-code relationship value for a code pair; and

generating the selected subset based at least in part on the first subset and the second subset.

8 . The computer-implemented method of claim 1 , wherein generating the selected subset comprises:

determining, based at least in part on the plurality of time bound code transition likelihood data objects, a plurality of inferred occurrence pathways of the plurality of individual historic code occurrences;

generating a plurality of per-pathway text representations corresponding to the plurality of inferred occurrence pathways;

determining, based at least in part on a sequence of the plurality of per-pathway text representations and using a language-based machine learning model, an inferred subset of the plurality of defined occurrence codes for a predictive entity; and

generating the selected subset based at least in part on the inferred subset.

9 . A system comprising:

one or more processors; and

at least one memory storing processor-executable instructions that, when executed by any one or more of the one or more processors, cause the one or more processors to perform operations comprising:

identifying a plurality of time bound code transition likelihood data objects for an entity cluster that is associated with an identifiable data entity, wherein: (i) a time bound code transition likelihood data object of the plurality of time bound code transition likelihood data objects is associated with a defined time bound of a plurality of defined time bounds, (ii) the time bound code transition likelihood data object for the defined time bound describes, for a code pair comprising a first defined occurrence code of a plurality of defined occurrence codes and a second defined occurrence code of the plurality of defined occurrence codes, an inferred likelihood that the second defined occurrence code occurs within the defined time bound of an assumed occurrence of the first defined occurrence code, and (iii) the time bound code transition likelihood data object is generated for the defined time bound by:

identifying a plurality of per-entity code occurrences for the entity cluster, wherein a per-entity code occurrence of the plurality of per-entity code occurrences describes that a corresponding defined occurrence code occurs in relation to a corresponding identifiable data entity in the entity cluster at a corresponding timestamp, and

generating the time bound code transition likelihood data object based at least in part on the plurality of per-entity code occurrences;

generating, based at least in part on the plurality of time bound code transition likelihood data objects and a plurality of individual historic code occurrences for the identifiable data entity, a future occurrence code prediction, for the identifiable data entity, comprising a selected subset of the plurality of defined occurrence codes; and

initiating one or more prediction-based actions based at least in part on the future occurrence code prediction.

10 . The system of claim 9 , wherein the operations further comprise:

determining an entity-code occurrence data object for the entity cluster based at least in part on the plurality of per-entity code occurrences; and

generating an augmented entity-code occurrence data object based at least in part on the entity-code occurrence data object.

11 . The system of claim 10 , wherein generating the augmented entity-code occurrence data object comprises:

performing singular value decomposition on the entity-code occurrence data object to generate a plurality of target lower-ranked entity-code occurrence data objects;

determining at least one missing value in the entity-code occurrence data object based at least in part on the plurality of target lower-ranked entity-code occurrence data objects; and

generating the augmented entity-code occurrence data object based at least in part on the at least one missing value.

12 . The system of claim 10 , wherein generating the time bound code transition likelihood data object based at least in part on the plurality of per-entity code occurrences comprises:

determining a plurality of augmented per-entity code occurrences for the entity cluster based at least in part on the plurality of per-entity code occurrences and the augmented entity-code occurrence data object;

determining a time bound code transition document that describes all co-occurrences of code pairs from the plurality of defined occurrence codes associated with the plurality of augmented per-entity code occurrences that occur within the defined time bound; and

generating the time bound code transition likelihood data object based at least in part on the time bound code transition document.

13 . The system of claim 12 , wherein generating the time bound code transition likelihood data object based at least in part on the time bound code transition document comprises:

determining a time bound code transition likelihood object for the code pair based at least in part on a per-document occurrence of the code pair as described by the time bound code transition document for the defined time bound and a cross-document occurrence of the code pair as described by the time bound code transition document; and

generating the time bound code transition likelihood data object based at least in part on the time bound code transition likelihood object.

14 . The system of claim 13 , wherein the time bound code transition likelihood object is a term-frequency-inverse-domain-frequency measure.

15 . The system of claim 9 , wherein generating the selected subset comprises:

generating a first subset of the plurality of defined occurrence codes based at least in part on the plurality of time bound code transition likelihood data objects and the plurality of individual historic code occurrences for the identifiable data entity;

generating a second subset of the plurality of defined occurrence codes based at least in part on a cross-code relationship value for a code pair; and

generating the selected subset based at least in part on the first subset and the second subset.

16 . The system of claim 9 , wherein generating the selected subset comprises:

determining, based at least in part on the plurality of time bound code transition likelihood data objects, a plurality of inferred occurrence pathways of the plurality of individual historic code occurrences;

generating a plurality of per-pathway text representations corresponding to the plurality of inferred occurrence pathways;

determining, based at least in part on a sequence of the plurality of per-pathway text representations and using a language-based machine learning model, an inferred subset of the plurality of defined occurrence codes for a predictive entity; and

generating the selected subset based at least in part on the inferred subset.

17 . One or more non-transitory computer-readable storage media including instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

identifying a plurality of time bound code transition likelihood data objects for an entity cluster that is associated with an identifiable data entity, wherein: (i) a time bound code transition likelihood data object of the plurality of time bound code transition likelihood data objects is associated with a defined time bound of a plurality of defined time bounds, (ii) the time bound code transition likelihood data object for the defined time bound describes, for a code pair comprising a first defined occurrence code of a plurality of defined occurrence codes and a second defined occurrence code of the plurality of defined occurrence codes, an inferred likelihood that the second defined occurrence code occurs within the defined time bound of an assumed occurrence of the first defined occurrence code, and (iii) the time bound code transition likelihood data object is generated for the defined time bound by:

identifying a plurality of per-entity code occurrences for the entity cluster, wherein a per-entity code occurrence of the plurality of per-entity code occurrences describes that a corresponding defined occurrence code occurs in relation to a corresponding identifiable data entity in the entity cluster at a corresponding timestamp, and

generating the time bound code transition likelihood data object based at least in part on the plurality of per-entity code occurrences;

generating, based at least in part on the plurality of time bound code transition likelihood data objects and a plurality of individual historic code occurrences for the identifiable data entity, a future occurrence code prediction, for the identifiable data entity, comprising a selected subset of the plurality of defined occurrence codes; and

initiating one or more prediction-based actions based at least in part on the future occurrence code prediction.

18 . The one or more non-transitory computer-readable storage media of claim 17 , wherein the operations further comprise:

determining an entity-code occurrence data object for the entity cluster based at least in part on the plurality of per-entity code occurrences; and

generating an augmented entity-code occurrence data object based at least in part on the entity-code occurrence data object.

19 . The one or more non-transitory computer-readable storage media of claim 18 , wherein generating the augmented entity-code occurrence data object comprises:

performing singular value decomposition on the entity-code occurrence data object to generate a plurality of target lower-ranked entity-code occurrence data objects;

determining at least one missing value in the entity-code occurrence data object based at least in part on the plurality of target lower-ranked entity-code occurrence data objects; and

generating the augmented entity-code occurrence data object based at least in part on the at least one missing value.

20 . The one or more non-transitory computer-readable storage media of claim 18 , wherein generating the time bound code transition likelihood data object based at least in part on the plurality of per-entity code occurrences comprises:

determining a plurality of augmented per-entity code occurrences for the entity cluster based at least in part on the plurality of per-entity code occurrences and the augmented entity-code occurrence data object;

determining a time bound code transition document that describes all co-occurrences of code pairs from the plurality of defined occurrence codes associated with the plurality of augmented per-entity code occurrences that occur within the defined time bound; and

generating the time bound code transition likelihood data object based at least in part on the time bound code transition document.

Assignments (1)

Continuity (1)

References Cited (44)