Granted Patent B2

US 11,544,652 · App. 14/829,607 · Granted Jan 3, 2023

Inventors: Vishnuvyas Sethumadhavan (Mountain View, CA); John O. Schneider (Los Gatos, CA); Richard Joseph Belcinski (San Jose, CA)

Assignee: Apixio, Inc.

G06Q10/0633G16H10/60G16H40/20G06Q10/10

Abstract

Claims (75)

1. In a health information management system, the health information management system comprising a processor in communication with a memory, a computer-implemented method for workflow management comprising:

receiving a plurality of records, the plurality of records including medical records, patient entered information, care team entered information, healthcare device generated information, and billing information;

determining a first subset of the plurality of records that include first machine readable objects and a second subset of the plurality of records that include machine unreadable objects;

converting, based at least in part upon optical character recognition (OCR), the second subset into an updated second subset including second machine readable objects;

embedding information in the first subset and the updated second subset by meta tagging the information into the first subset and the second subset to characterize the first subset and the updated second subset;

extracting, by a reconciliation engine utilizing a set of similarity rules for determining which data to cluster into a plurality of clusters and based at least in part upon the embedded information, a plurality of data elements from the first subset and the updated second subset;

transmitting the plurality of data elements to a concept model, wherein the concept model is configured to:

receive the plurality of data elements from the reconciliation engine;

receive a query from a user;

associate, based at least in part upon the query, each of the plurality of data elements with a respective cluster of the plurality of clusters by utilizing a set of clustering rules, thereby mapping each of the plurality of data elements based upon similarities and medical concepts;

apply a set of dynamic time rules to each of the plurality of data elements, wherein the dynamic time rules represent changes in patient data over time;

dynamically update the set of clustering rules and the set of dynamic time rules based upon feedback;

determine, based upon the query, a threshold distance between each of the plurality of data elements and the respective cluster;

determine, based upon the mapping and the threshold distance, which data elements of the plurality of data elements satisfy the threshold distance with respect to the respective cluster; and

bundle each of the data elements that satisfy the threshold distance into a plurality of work items and reject each of the data elements that do not satisfy the threshold distance;

configuring at least one workflow based upon event history for the work items, current action, and user context;

receiving a workflow objective comprising maximizing reimbursement of MediCare claims;

determining, for each of the plurality of work items, a probability that a next action taken in the at least one workflow for each particular work item will increase at least one of an acceptance probability by MediCare for reimbursement and a reimbursement amount; and

routing the work items through the at least one workflow, wherein work items associated with a higher probability of acceptance and higher reimbursement amounts are routed before work items with a lower probability of acceptance and lower reimbursement amounts in order to achieve the workflow objective of maximizing reimbursement of MediCare claims.

2. The method of claim 1 wherein the work items are bundles of data elements for which value is to be extracted for a particular user objective.

3. The method of claim 1 wherein the routing is either exploration or exploitation based upon confidence in a probabilistic model for workflow action outcomes.

4. The method of claim 3 wherein the exploration routing collects results from processing of work items through a given workflow action in order to tune the probabilistic model.

5. The method of claim 3 wherein the exploitation routing compares an expected outcome for the processing of each work item through the workflow versus a user objective and selects work items that maximize the objective.

6. The method of claim 1 further comprising updating the workflow configuration based upon the results of an action.

7. The method of claim 1 wherein the bundling comprises:

selecting an attribute for clustering;

define criteria for a cluster based upon a user objective;

computing the distance between the attribute for a given data element, and the cluster; and

including data elements for which the distance is below a threshold and rejecting data elements for which the distance is above the threshold.

8. The method of claim 1 further comprising sampling the work items in a given workflow for quality control.

9. The method of claim 1 further comprising:

parsing the medical information;

normalizing the medical information;

indexing the medical information to generate an initial set of data elements;

tagging the initial set of data elements; and

iteratively processing the data elements to generate the plurality of data elements, wherein the processing includes at least one of first order logic querying and knowledge extraction to infer at least one additional element which is added to the plurality of data elements.

10. A health information management system for workflow management comprising:

an interface for receiving a query from a user;

a bundler comprising a processor, the bundler configured to map each of a plurality of data elements to a medical concept according to a concept model, wherein the bundler is configured to:

receive a plurality of records, the plurality of records including medical records, patient entered information, care team entered information, healthcare device generated information, and billing information;

determine a first subset of the plurality of records that include first machine readable objects and a second subset of the plurality of records that include machine unreadable objects;

convert, based at least in part upon optical character recognition (OCR), the second subset into an updated second subset including second machine readable objects;

embed information in the first subset and the updated second subset by meta tagging the information into the first subset and the second subset to characterize the first subset and the updated second subset;

extract, by a reconciliation engine utilizing a set of similarity rules for determining which data to cluster into a plurality of clusters and based at least in part upon the embedded information, a plurality of data elements from the first subset and the updated second subset; and

transmit the plurality of data elements to the concept model, wherein the concept model is configured to:

receive the plurality of data elements from the reconciliation engine;

receive the query from the user;

associate, based at least in part upon the query, each of the plurality of data elements with a respective cluster of the plurality of clusters by utilizing a set of clustering rules, thereby mapping each of the plurality of data elements based upon similarities and medical concepts;

apply a set of dynamic time rules to each of the plurality of data elements, wherein the dynamic time rules represent changes in patient data over time;

dynamically update the set of clustering rules and the set of dynamic time rules based upon feedback;

determine, based upon the query, a threshold distance between each of the plurality of data elements and the respective cluster;

determine, based upon the mapping and the threshold distance, which data elements of the plurality of data elements satisfy the threshold distance with respect to the respective cluster; and

bundle each of the data elements that satisfy the threshold distance into a plurality of work items and reject each of the data elements that do not satisfy the threshold distance;

at least one state machine comprising a processor, the at least one state machine configured to contextually configure at least one workflow based upon event history for the work items, current action, and user context; and

an energy based router comprising a processor, the energy based router configured to:

receive a workflow objective comprising maximizing reimbursement of MediCare claims;

determine, for each of the plurality of work items, a probability that a next action taken in the at least one workflow for each particular work item will increase at least one of an acceptance probability by MediCare for reimbursement and a reimbursement amount; and

route the work items through the at least one workflow, wherein work items associated with a higher probability of acceptance and higher reimbursement amounts are routed before work items with a lower probability of acceptance and lower reimbursement amounts in order to achieve the workflow objective of maximizing reimbursement of MediCare claims.

11. The system of claim 10 wherein the work items are bundles of data elements for which value is to be extracted for a particular user objective.

12. The system of claim 10 wherein the energy based router operates in either exploration or exploitation mode based upon confidence in a probabilistic model for workflow action outcomes.

13. The system of claim 12 wherein the exploration mode of the energy based router collects results from processing of work items through a given workflow action in order to tune the probabilistic model.

14. The system of claim 12 wherein the exploitation mode of the energy based router compares an expected outcome for the processing of each work item through the workflow versus a user objective and selects work items that maximize the objective.

15. The system of claim 10 wherein the at least one state machine is further configured to update the workflow configuration based upon the results of an action.

16. The system of claim 10 wherein the bundler is configured to:

select an attribute for clustering;

define criteria for a cluster based upon a user objective;

compute the distance between the attribute for a given data element, and the cluster; and

include data elements for which the distance is below a threshold and reject data elements for which the distance is above the threshold.

17. The system of claim 10 further comprising a quality checker configured to sample the work items in a given workflow for quality control.

18. The system of claim 10 further comprising a contextual analyzer configured to:

parse the medical information;

normalize the medical information;

index the medical information to generate an initial set of data elements;

tag the initial set of data elements; and

iteratively process the data elements to generate the plurality of data elements, wherein the processing includes at least one of first order logic querying and knowledge extraction to infer at least one additional element which is added to the plurality of data elements.

Assignments (5)

Continuity (11)