Deep semantic search of electronic medical records

Patent metadata

Field	Value
Publication number	US-9690861-B2
Application number	US-201414334033-A
Country	US
Kind code	B2
Filing date	Jul 17, 2014
Priority date	Jul 17, 2014
Publication date	Jun 27, 2017
Grant date	Jun 27, 2017

Abstract

Methods, systems, and devices provide semantically relevant information by analyzing an Electronic Medical Record (EMR) having structured data and unstructured data. In the analysis, a first set of medical concepts is identified from the unstructured data in the EMR, and a second set of medical concepts is identified from the structured data in the EMR. Relationships between medical concepts in the first set of medical concepts and the second set of medical concepts are automatically identified in a medical ontology by such methods, systems, and devices.

First claim

What is claimed is:

1. A method comprising:
receiving a query for information from an electronic medical record (EMR) comprising structured data and unstructured data;
annotating contents of said unstructured data and said structured data to produce annotations;
using said annotations to create concept unique identifiers (CUIs);
identifying clinically relevant semantic relationships between said structured data and unstructured data in said EMR based on statistical associations between said CUIs;
producing a score for relevant information from said EMR that is semantically related to said query based on strength of said clinically relevant semantic relationships between said structured data and unstructured data; and
prioritizing a display of said relevant information based on said score; and
providing, in response to said query, said relevant information, said relevant information comprising at least one of clinical notes, medications, test results, treatments, and contraindications.

2. The method according to claim 1 , further comprising:
determining whether EMR data is semantically relevant to said query based on medical relationship relevancy.

3. The method according to claim 1 , further comprising:
identifying a first set of medical concepts from said unstructured data in said EMR;
identifying a second set of medical concepts from said structured data in said EMR; and
identifying clinically relevant semantic relationships in a medical ontology between medical concepts in said first set of medical concepts and said second set of medical concepts.

4. The method according to claim 3 , further comprising:
creating inverted search indexes on said first set of medical concepts and said second set of medical concepts, said inverted search indexes enabling retrieval of passages matching said query syntactically and semantically.

5. The method according to claim 3 , said identifying clinically relevant semantic relationships in said medical ontology between medical concepts in said first set of medical concepts and said second set of medical concepts further comprising:
identifying causation of medical conditions and treatments for medical conditions based on said medical concepts.

6. The method according to claim 1 , further comprising:
outputting results of said query in textual form to a user interface.

7. The method according to claim 1 , said CUIs comprise standardized identifiers relating to medical disorders related to said information in said unstructured data and said structured data.

8. A method comprising:
analyzing an electronic medical record (EMR) comprising structured data and unstructured data;
annotating contents of said unstructured data and said structured data to produce annotations;
using said annotations to create concept unique identifiers (CUIs), said analyzing comprising:
automatically identifying a first set of medical concepts from said unstructured data in said EMR;
automatically identifying a second set of medical concepts from said structured data in said EMR; and
automatically identifying clinically relevant semantic relationships in a medical ontology between medical concepts in said first set of medical concepts and said second set of medical concepts in said EMR based on statistical associations between said CUIs;

producing a score for relevant information from said EMR that is semantically related to a query based on strength of said clinically relevant semantic relationships between said first set of medical concepts and said second set of medical concepts;
prioritizing a display of said relevant information based on said score;
storing relationship information as a data structure in a computerized device; and
providing, in response to said query, said relevant information.

9. The method according to claim 8 , said automatically identifying relationships in a medical ontology between medical concepts in said first set of medical concepts and said second set of medical concepts further comprising:
identifying causation of medical conditions and treatments for medical conditions based on said medical concepts.

10. The method according to claim 8 , further comprising:
receiving a query containing search terms for information from said EMR; and
retrieving semantically relevant results from said EMR related to said search terms in response to said query using a search index.

11. The method according to claim 10 , further comprising:
outputting retrieved semantically relevant results.

12. The method according to claim 8 , said CUIs comprise standardized identifiers relating to medical disorders related to said information in said unstructured data and said structured data.

13. A system comprising:
a storage system storing electronic medical records (EMRs) comprising structured data and unstructured data;
an I/O interface configured to receive a query for information from an EMR; and
a processing unit,
said processing unit being configured to annotate contents of said unstructured data and said structured data to produce annotations,
said processing unit being configured to use said annotations to create concept unique identifiers (CUIs),
said processing unit being configured to identify clinically relevant semantic relationships between said structured data and unstructured data in said EMR based on statistical associations between said CUIs,
said processing unit being configured to produce a score for relevant information from said EMR that is semantically related to said query based on strength of said clinically relevant semantic relationships between said structured data and unstructured data,
said processing unit being configured to prioritize a display of said relevant information based on said score, and
said I/O interface begin configured to generate results to said query based on said relevant information, wherein said results comprise at least one of clinical notes, medications, test results, treatments, and contraindications.

14. The system according to claim 13 , said CUIs comprise standardized identifiers relating to medical disorders related to said information in said unstructured data and said structured data.

15. A computer program product for creating a semantically searchable electronic medical record, said computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions being readable/executable by a processor, to cause said processor to perform a method comprising:
receiving a query for information from an electronic medical record (EMR) comprising structured data and unstructured data;
annotating contents of said unstructured data and said structured data to produce annotations;
using said annotations to create concept unique identifiers (CUIs);
identifying clinically relevant semantic relationships between said structured data and unstructured data in said EMR based on statistical associations between said CUIs;
producing a score for relevant information from said EMR that is semantically related to said query based on strength of said clinically relevant semantic relationships between said structured data and unstructured data;
prioritizing a display of said relevant information based on said score; and
providing, in response to said query, said relevant information,
said relevant information comprising at least one of clinical notes, medications, test results, treatments, and contraindications.

16. The computer program product according to claim 15 , said method further comprising:
determining wheth

Assignees

• IBM

Inventors

• Boloor Keerthana
• Brown Eric W
• Devarakonda Murthy V
• Ferrucci David
• Prager John M

Classifications

• G16H10/60

  for patient-specific data, e.g. for electronic patient records · CPC title

• G06F16/36

  Creation of semantic tools, e.g. ontology or thesauri · CPC title

• G06F16/9535 Primary

  Search customisation based on user profiles and personalisation · CPC title

• G06F16/367 Primary

  Ontology · CPC title

• G06F19/322

  Physics · mapped topic