MACHINE LEARNING CLINICAL DECISION SUPPORT SYSTEM FOR RISK CATEGORIZATION

• US 20170140114A1
• Filed: 11/14/2016
• Published: 05/18/2017
• Est. Priority Date: 11/13/2015
• Status: Active Grant

• Current Assignee: Cerner Innovation, Inc. (Oracle Corporation)
• Original Assignee: Cerner Innovation, Inc. (Oracle Corporation)
• Inventors: ARE, SASANKA, SUGDEN, JOHN MATTHEW
• Granted Patent: US 10,650,927 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

Improved risk categorization is provided for clinical decision support and forecasting future health care spend. A risk index is provided that improves on other risk stratification models by synthesizing electronic medical records and health questionnaires with an individual patient'"'"'s claim histories. Machine learning algorithms catalogue patients into distinct group clusters, based on risk which may be associated with annual health care spending, thereby enabling administrators to forecast future health care spending on the individual and population level.

Claims

1. A computerized system comprising:

a risk index classifier configured to classify members of a population of humans into one or more risk groups from a set of risk groups;

one or more processors; and

computer storage memory having computer-executable instructions stored thereon which, when executed by the processor, implement a method of predicting likely future health care spend for the population members, the method comprising;

receiving a set of member data for a plurality of members of the population;

determining one or more risk index models used by the risk index classifier, the one or more risk index models determined using a set of machine learning algorithms;

classifying, using the risk index classifier, one or more members of the population into one or more risk groups from the set of risk groups, based on the set of member data;

determining a future health care spend forecast for the classified one or more members of the population; and

modifying a health care event based on the determined forecast,

wherein the set of member data comprises data related to claims and data related to at least one of (i) wellness or personal-health-assessment (PHA) information and (ii) electronic medical record (EMR) information.

2. The computerized system of claim 1, wherein the modified health care event comprises selecting or modifying a computer health care treatment program associated with the classified one or more members of the population.

3. The computerized system of claim 1, wherein the set of member data comprises data related to claims, wellness or personal-health-assessment (PHA) information and electronic medical record (EMR) information.

4. The computerized system of claim 1 further comprising:

aggregating the receiving a set of member data; and

determining interpretative data based on the aggregated set of member data,

wherein classifying the one or more members of the population into one or more risk groups from the set of risk groups is further based on the interpretative data.

5. The computerized system of claim 1, wherein determining a future health care spend forecast is based at least in part on the received set of member data.

6. The computerized system of claim 1, wherein one or more index models are determined at least based on the received set of member data.

7. The computerized system of claim 1, wherein the set of machine learning algorithms comprises an ensemble of alternating decision trees.

8. The computerized system of claim 7 wherein the set of machine learning algorithms further comprises a cost-sensitive classifier, a multi-class classifier, and a rotational forest ensemble method.

9. The computerized system of claim 1 wherein the set of machine learning algorithms comprises a cost-sensitive classifier.

10. The computerized system of claim 1, wherein determining the one or more risk index models comprises selecting one or more pre-determined risk index models and updating the selected one or more risk index models based at least in part on the received set of member data.

11. The computerized system of claim 1 further comprising determining the number of risk groups in the set of risk groups, based in part on the received set of member data.

12. The computerized system of claim 1, wherein the set of member data comprises data related to claims and behavior or lifestyle information.

13. A method for predicting likely future health care spend for individual members of a population of humans using a computing system, the method comprising:

receiving electronic data over a communications network for the members of the population, the data comprising historical information related to claims and data related to at least one of (i) wellness or personal-health-assessment (PHA) information and (ii) electronic medical record (EMR) information;

classifying, via the computing system, members of the population into a set of risk groups, based on the received data and a risk index model, the risk index model using one or more machine learning algorithms; and

determining, via a computing system, a set of future health care costs over a duration of time associated with the classified members based on the classification and the received data.

14. The method of claim 13 further comprising modifying a health care event based on the determined set of future health care costs.

15. The method of claim 13, wherein the one or more machine learning algorithms comprises an ensemble of alternating decision trees.

16. The method of claim 15, wherein the set of machine learning algorithms further comprises a cost-sensitive classifier.

17. The method of claim 13 further comprising determining a future health care cost the over a duration of time for the population based on the set of future health care costs associated with the classified members.

18. One or more computer-storage media having computer-executable instructions embodied thereon that, when executed by a computing device, cause the computing device to perform a method of determining future health care costs for members in a population of human patients, the method comprising:

receiving raw data for the members in the population, the raw data comprising claims information and at least one of (i) wellness or personal-health-assessment (PHA) information and (ii) electronic medical record (EMR) information for each of the members;

determining interpretive data based on the raw data;

combining the raw data and interpretive data thereby forming composite data;

determining a risk index model based on the composite data and a set of machine learning algorithms;

classifying the members of the population into one or more risk groups using the risk index model;

determining a future health care costs over a duration of time for the members based on the risk groups for which the members are classified; and

modifying a health care event based on the determined future health care cost.

19. The media claim 18, wherein the set of machine learning algorithms comprises an ensemble of alternating decision trees.

20. The media claim 18, wherein the set of machine learning algorithms comprises a cost-sensitive classifier.