AUTOMATED MANAGEMENT OF MEDICAL DATA USING EXPERT KNOWLEDGE AND APPLIED COMPLEXITY SCIENCE FOR RISK ASSESSMENT AND DIAGNOSES

• US 20100094648A1
• Filed: 10/13/2009
• Published: 04/15/2010
• Est. Priority Date: 10/10/2008
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: Cardiovascular Imaging Technologies LLC
• Inventors: Seward, James B.
• Granted Patent: US 8,554,580 B2
• Time in Patent Office: Days
• US Class Current: 705/2
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

A knowledgebase comprising a plurality of feature-sets is created using complexity science. The knowledgebase is accessible through a network to a computer, a system, and a computer program code that receives and analyzes medical data to determine the existence of a disease state. The medical data is input and correlated to features within the knowledgebase to identify feature-sets, each feature-set indicating a particular medical condition. After one or more feature-sets have been selected, associative algorithms consider the magnitudes or values of the medical data and, with the values of features in the feature-sets, assess the risk burden of the medical condition associated with the feature-set. An output is generated that may include diagnoses of one or more medical conditions, the risk burden of the medical condition(s), possible treatment options and prevention techniques.

Claims

1. A method for evaluating medical data of a person to identify, if any, at-risk medical conditions, comprising:

obtaining with a computer the medical data, said medical data having features of at least one of various medical conditions, at least some of the features of said medical data having values;

accessing from a memory a medical knowledgebase having a plurality of feature-sets relating to the various medical conditions, each of the plurality of feature-sets having a group of highly-associated features relating to particular ones of the various medical conditions, at least some of said highly-associated features having ranges of values;

determining with the computer a subset of the plurality of feature-sets by correlating at least two of the features of the medical data with at least two of said highly-associated features of each of the feature-sets in the subset thereby transforming knowledge of the features of the medical data to metadata in the form of the group of transformed highly-associated features of each of the feature-sets in the subset;

comparing with the computer whether the features of the medical data are one of normal or abnormal and magnitudes of values of the medical data are within the ranges of values of the transformed highly-associated features which are normal or abnormal to interpret relative to a standard so as to identify, if any, at-risk medical conditions of said person, and

outputting from the computer information relating to the at-risk medical condition.

2. The method of claim 1, including outputting from the computer a request to acquire additional medical data to further assess the at-risk medical conditions.

3. The method of claim 1, wherein the ranges of values of said highly-associated features are each a plurality of ranges of values such that a risk level is assigned for each different range of values of the plurality of ranges of values, said computer calculating a risk level for the at-risk medical condition based on the risk levels of the transformed highly-associated features of the feature-sets due to the values of the medical data.

4. The method of claim 1, including identifying an intensity of association levels of the transformed highly-associated features of the feature-sets with a state of the at-risk medical condition based on a magnitude of the values of the features of the medical data, said state being one of none, pre-emergent, emerging, and expressed.

5. The method of claim 1, wherein the at-risk medical condition is identified from no more than 4 transformed highly-associated features for a particular one of said feature-sets in the subset.

6. A computer system for evaluating medical data to identify an at-risk medical condition, said computer system comprising a central processing unit coupled to a memory, said central processing unit being programmed to evaluate the medical data by:

obtaining the medical data, said medical data having features of at least one of various medical conditions, at least some of the features of said medical data having values;

accessing from the memory a medical knowledgebase having a plurality of feature-sets relating to the various medical conditions, each of the plurality of feature-sets having a group of highly-associated features relating to particular ones of the various medical conditions, at least some of said highly-associated features having ranges of values;

determining with the central processing unit a subset of the plurality of feature-sets by correlating at least two of the features of the medical data with at least two of said highly-associated features of each of the feature-sets in the subset thereby transforming knowledge of the features of the medical data to metadata in the form of the group of transformed highly-associated features of each of the feature-sets in the subset;

comparing with the central processing unit whether the features of the medical data are one of normal or abnormal and magnitudes of values of the medical data are within the ranges of values of the transformed highly-associated features which are normal or abnormal to interpret relative to a standard so as to identify, if any, at-risk medical conditions of said person; and

outputting from an interface information relating to the at-risk medical condition.

7. The system of claim 6, wherein the ranges of values of said highly-associated features are each a plurality of ranges of values, said central processing unit being further programmed to assign a risk level for each different range of values of the plurality of ranges of values.

8. The system of claim 6, including said central processing unit being further programmed to identify an intensity of association levels of the transformed highly-associated features of the feature-sets with a state of the at-risk medical condition based on a magnitude of the values of the features of the medical data, said state being one of none, pre-emergent, emerging, and expressed.

9. A non-transitory computer-readable storage medium with an executable program stored thereon, said program evaluating medical data, said medium loaded on a computer system wherein said program instructs a central processing unit coupled to a memory and to a data receiving interface to perform the following steps:

obtain the medical data, said medical data having features of at least one of various medical conditions, at least some of the features of said medical data having values;

access from the memory a medical knowledgebase having a plurality of feature-sets relating to the various medical conditions, each of the plurality of feature-sets having a group of highly-associated features relating to particular ones of the various medical conditions, at least some of said highly-associated features having ranges of values;

determine a subset of the plurality of feature-sets by correlating at least two of the features of the medical data with at least two of said highly-associated features of each of the feature-sets in the subset thereby transforming knowledge of the features of the medical data to metadata in the form of the group of transformed highly-associated features of each of the feature-sets in the subset;

comparing whether the features of the medical data are one of normal or abnormal and magnitudes of values of the medical data are within the ranges of values of the transformed highly-associated features which are normal or abnormal to interpret relative to a standard so as to identify, if any, at-risk medical conditions of said person; and

output from an interface information relating to the at-risk medical condition.

10. The storage medium of claim 9, wherein the ranges of values of the features are each a plurality of ranges of values, said program including a further step of assigning a risk level for each different range of values of the plurality of ranges of values.

11. The storage medium of claim 9, wherein said program includes a further step of identifying an intensity of association levels of the transformed highly-associated features of the feature-sets with a state of the at-risk medical condition based on a magnitude of the values of the features of the medical data, said state being one of none, pre-emergent, emerging, and expressed.

12. A method of adding or modifying a candidate feature-set relating to a medical condition for a medical knowledgebase including feature-sets relating to various medical conditions wherein medical data of a person is evaluated relative to the feature-sets to identify an at-risk medical condition, at least some of the medical data having values, said medical data having a feature of at least one of various medical conditions, said method comprising:

inputting to a computer at least one candidate feature to be considered relative to the candidate feature-set, said candidate feature-set having at least one other existing feature;

comparing with the computer the at least one candidate feature with the at least one other existing feature;

selecting the at least one candidate feature for inclusion in the candidate feature-set if when the at least one candidate feature is one of abnormal and within a range of values which is abnormal, there is a correlative effect with the at least one other existing feature such that together they have an increased association level with the medical condition to which the candidate feature-set relates; and

including the candidate feature-set in the medical knowledgebase.

13. The method of claim 12, wherein the candidate feature and the at least one other existing feature have ranges of values, at least one of said ranges of values being a plurality of ranges of values such that a risk level is assigned for each different range of values of the plurality of ranges of values.

14. The method of claim 12, including correlating a magnitude of a predetermined association level of the plurality of features of the candidate feature-set with a state of the at-risk medical condition, said state being one of none, pre-emergent, emerging, and expressed.