Medical ontologies for computer assisted clinical decision support

• US 7,630,947 B2
• Filed: 08/16/2006
• Issued: 12/08/2009
• Est. Priority Date: 08/25/2005
• Status: Active Grant

• Current Assignee: Cerner Innovation, Inc. (Oracle Corporation)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Steck, Harald, Rosales, Romer E., Rao, R. Bharat, Pandya, Abhinay Mahesh
• Primary Examiner(s): Holmes; Michael B
• Application Number:
  US11/505,080
• Publication Number:
  US 20070094188A1
• Time in Patent Office: 1,210 Days
• Field of Search: 706/45
• US Class Current: 706/45
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

Medical ontology information is used for mining and/or probabilistic modeling. A domain knowledge base may be automatically or semi-automatically created by a processor from a medical ontology. The domain knowledge base, such as a list of disease associated terms, is used to mine for corresponding information from a medical record. The relationship of different terms with respect to a disease may be used to train a probabilistic model. Probabilities of a disease or chance of indicating the disease are determined based on the terms from a medical ontology. This probabilistic reasoning is learned with a machine from ontology information and a training data set.

Claims

1. A method for use of a medical ontology for computer assisted clinical decision support, the method comprising:

identifying, with a processor of a data mining system, a plurality of associated terms from a medical ontology;

generating, with the processor of the data mining system, a domain-knowledge base from the associated terms; and

mining, with the data mining system, a medical record as a function of the domain-knowledge base.

2. The method of claim 1 wherein identifying comprises identifying the associated terms as having a IS-A type relationship from the medical ontology.

3. The method of claim 1 wherein identifying comprises identifying at least some of the associated terms as causes, effects, symptoms, signs, related diseases, body locations, morphology, or combinations thereof of a disease.

4. The method of claim 1 wherein identifying comprises identifying at least some of the associated terms of a clinical concept, the clinical concept other than a disease.

5. The method of claim 1 wherein identifying comprises identifying the associated terms from the medical ontology and an additional medical ontology.

6. The method of claim 1 wherein generating comprises assigning a variable to each of the associated terms.

7. The method of claim 1 further comprising:

restricting the identification of associated terms as a function of distance in the medical ontology, type of association, or combinations thereof.

8. The method of claim 1 wherein mining the medical record comprises searching for the associated term or terms in structured and unstructured portions of the medical record.

9. The method of claim 1 wherein mining the medical record comprises inferring a patient state as a function of the associated terms and corresponding probabilities of the associated terms indicating the patient state.

10. The method of claim 1 further comprising:

training, with the processor, a graphical model as a function of the domain-knowledge base and training data.

11. The method of claim 1 further comprising:

receiving an update of the medical ontology; and

repeating the identifying and generating for the update.

12. A system for use of a medical ontology for computer assisted clinical decision support, the system comprising:

a memory operable to store a mining engine with a domain-knowledge base from associated terms in a medical ontology, the mining engine operable to mine a medical record as a function of the domain-knowledge base; and

a processor of a mining system operable to apply the mining engine to the medical record.

13. The system of claim 12 wherein the associated terms from the medical ontology are from three or more of the groups of:

having a IS-A type relationship, a cause, an effect, a symptom, a sign, a related disease, a body location, and morphology.

14. The system of claim 12 wherein a variable is assigned to each of the associated terms, and wherein the mining engine is operable to search for the associated terms in structured and unstructured portions of the medical record and infer a patient state as a function of the associated terms and corresponding probabilities of the associated terms indicating the patient state.

15. In a computer readable storage media having stored therein data representing instructions executable by a programmed processor for use of a medical ontology for computer assisted clinical decision support, the storage media comprising instructions for:

creating a knowledge base from a medical ontology; and

building a mining engine for a mining system and operable to search unstructured medical data as a function of the knowledge base.

16. The instructions of claim 15 wherein creating the knowledge base comprises identifying a plurality of associated terms from the medical ontology and generating the knowledge base from the associated terms.

17. A method for use of a medical ontology for computer assisted clinical decision support, the method comprising:

identifying a plurality of associated terms from a medical ontology; and

training a graphical model of relationships and probabilities of the associated terms as a function of the associated terms and training data of medical records representing patients, the graphical model operable for computer assisted clinical decision support.

18. The method of claim 17 wherein identifying comprises identifying the associated terms as having a IS-A type relationship from the medical ontology.

19. The method of claim 17 wherein identifying comprises identifying at least some of the associated terms as causes, effects, symptoms, signs, related diseases, body locations, morphology, or combinations thereof of a disease.

20. The method of claim 17 wherein training comprises assigning a variable in the graphical model to each associated term.

21. The method of claim 20 wherein training comprises constructing the training data as a plurality of vectors, each vector having a location for each of the variables, labeling a disease of interest corresponding to the associated terms, and training the graphical model from the training data where at least one variable of at least one vector has a missing value.

22. The method of claim 20 further comprising:

creating a knowledge base from a medical ontology;

wherein training comprises building a mining engine operable to search unstructured medical data as a function of the knowledge base.

23. A system for use of a medical ontology for computer assisted clinical decision support, the system comprising:

a memory operable to store a probabilistic model having machine learned probabilities for relationships from a medical ontology, the probabilities machine learned from medical records representing patients; and

a processor operable to apply the probabilistic model for computer assisted clinical decision support.

24. The system of claim 23 wherein the relationships from the medical ontology comprise cause, effect, or combinations thereof for a disease.

25. The system of claim 23 wherein the relationships from the medical ontology comprise a different disease relationship with a disease of interest.

26. The system of claim 23 wherein the processor is operable to mine unstructured data of a medical record as a function of a knowledge base derived from the medical ontology, and wherein the processor is operable to apply the probabilistic model to results of the mining.

27. In a computer readable storage media having stored therein data representing instructions executable by a programmed processor for use of a medical ontology for computer assisted clinical decision support, the storage media comprising instructions for:

obtaining a medical record representing at least one patient;

determining, from the medical record, a chance of a disease as a function of IS A relationships and other relationships from a medical ontology; and

displaying the chance as an output of computer assisted clinical decision support.

28. The instructions of claim 27 wherein determining comprises applying a graphical model having machine learned probabilities associated with the IS A and other relationships.

29. The instructions of claim 27 further comprising:

mining the medical record as a function of a knowledge base derived from the medical ontology;

wherein determining comprises determining from an output of the mining.