Information theory entropy reduction program

• US 8,756,234 B1
• Filed: 11/16/2004
• Issued: 06/17/2014
• Est. Priority Date: 11/16/2004
• Status: Active Grant

• Current Assignee: The General Hospital Corporation (Mass General Brigham, Inc.)
• Original Assignee: The General Hospital Corporation (Mass General Brigham, Inc.)
• Inventors: Dreyer, Keith J.
• Primary Examiner(s): MORRISON, JAY A
• Application Number:
  US10/989,598
• Time in Patent Office: 3,500 Days
• Field of Search: 707/6, 705/2, 705/3
• US Class Current: 707/739
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

A system and process for automated text analysis which can be used to identify phrases in reports such as medical reports includes identifying a phrase contained within a text, extracting the phrase from the text, determining a value of the phrase and, in response to the phrase having at least a threshold value, reducing the phrase to a root meaning. In one embodiment, the value of the phrase is assigned via lexicon-based hierarchical decision trees.

Claims

1. A computer method, comprising:

processing automatically medical text in a medical report within a medical field using a computer, the processing comprising;

identifying medical phrases contained within the medical text;

extracting the medical phrases from the medical text;

determining which of the medical phrases are valued medical phrases, a valued medical phrase includes content having a corresponding value in a lexicon of relevant terminology corresponding to the medical field, the valued medical phrase being either a medical finding or a medical recommendation;

reducing each valued medical phrase to one or more medical codes in the medical field;

the one or more medical codes being medical identifiers; and

rendering in a first group medical codes corresponding to the medical finding; and

rendering in a second group separate from the first group medical codes corresponding to the medical recommendation,

wherein processing automatically the medical text in the medical report further comprises generating a lexicon-based hierarchical decision tree for the medical field by assigning each valued medical phrase to a location in the lexicon-based hierarchical decision tree.

2. The computer method of claim 1, wherein reducing each valued medical phrase to one or more medical codes in the medical field comprises reducing each valued medical phrase to one or more standard medical codes in the medical field.

3. The computer method of claim 1, wherein determining which of the medical phrases are valued medical phrases comprises determining which of the medical phrases are valued medical phrases including content having a corresponding value in a previously trained lexicon of relevant terminology corresponding to the medical field.

4. The computer method of claim 1 wherein rendering in the first group medical codes corresponding to the medical finding comprises displaying on a screen the first group,

wherein rendering in the second group medical codes corresponding to the medical recommendation comprises displaying on the screen the second group.

5. The computer method of claim 1, wherein determining which of the medical phrases are valued medical phrases comprises determining which of the medical phrases are valued medical phrases including content having a corresponding value in a lexicon of radiological terms.

6. The computer method of claim 1, wherein processing automatically the medical text within the medical field using the computer comprises processing automatically the medical text within radiology.

7. The computer method of claim 1, wherein reducing each valued medical phrase to one or more medical codes in the medical field comprises reducing each valued medical phrase to one or more radiology codes in radiology.

8. The computer method of claim 1, wherein reducing each valued medical phrase to one or more medical codes in the medical field comprises reducing each valued medical phrase to one or more codes from at least one of Current Procedural Terminology, Fourth Revision (CPT4), International Classification of Diseases, Ninth Revision (ICD9), Systematized Nomenclature of Medicine (SNOWMED), a lexicon of radiological terms and Unified Medical Language System (UMLS).

9. The computer method of claim 1, further comprising transcribing dictated words into the medical text using a voice recognition and transcription system.

10. An apparatus, comprising:

hardware circuitry configured to process automatically medical text in a medical report within a medical field, the circuitry configured to process automatically medical text comprises circuitry configured to;

identify medical phrases contained within the medical text;

extract the medical phrases from the medical text;

determine which of the medical phrases are valued medical phrases, a valued medical phrase includes content having a corresponding value in a lexicon of relevant terminology corresponding to the medical field, the valued medical phrase being either a medical finding or a medical recommendation;

reduce each valued medical phrase to one or more medical codes in the medical field;

the one or more medical codes being medical identifiers; and

render in a first group medical codes corresponding to the medical finding; and

render in a second group separate from the first group medical codes corresponding to the medical recommendation,

wherein the circuitry comprises a processor, and

wherein the circuitry configured to process automatically the medical text in the medical report further comprises circuitry configured to generate a lexicon-based hierarchical decision tree for the medical field by assigning each valued medical phrase to a location in the lexicon-based hierarchical decision tree.

11. The apparatus of claim 10, wherein the one or more medical codes in the medical field comprises one or more standard medical codes in the medical field.

12. The apparatus of claim 10, wherein the trained lexicon of relevant terminology corresponding to the medical field comprises a previously trained lexicon of relevant terminology corresponding to the medical field.

13. The apparatus of claim 11, wherein the circuitry configured to render in the first group medical codes corresponding to the medical finding comprises circuitry configured to display on a screen the first group;

and

wherein the circuitry configured to render in the second group medical codes corresponding to the medical recommendation comprises circuitry configured to display on the screen the second group.

14. The apparatus of claim 11, wherein the a lexicon of relevant terminology comprises a lexicon of radiological terms.

15. The apparatus of claim 11, wherein the medical field is radiology.

16. The apparatus of claim 11, wherein the one or more medical codes in the medical field comprises one or more radiology codes in radiology.

17. The apparatus of claim 11, wherein the one or more medical codes in the medical field comprises one or more codes from at least one of Current Procedural Terminology, Fourth Revision (CPT4), International Classification of Diseases, Ninth Revision (ICD9), Systematized Nomenclature of Medicine (SNOWMED), a lexicon of radiological terms, or Unified Medical Language System (UMLS).

18. The apparatus of claim 11, further comprising circuitry configured to transcribe dictated words into the medical text.

19. A computer method, comprising:

processing automatically medical text in a radiology report using a computer, the processing comprising;

identifying medical phrases contained within the medical text;

extracting the medical phrases from the medical text;

determining which of the medical phrases are valued medical phrases, a valued medical phrase includes content having a corresponding value in a lexicon of relevant terminology in radiology, the valued medical phrase being either a medical finding or a medical recommendation;

reducing each valued medical phrase to one or more radiology codes in radiology from at least one of Current Procedural Terminology, Fourth Revision (CPT4), International Classification of Diseases, Ninth Revision (ICD9), Systematized Nomenclature of Medicine (SNOWMED), a lexicon of radiological terms, or Unified Medical Language System (UMLS);

the one or more medical codes being medical identifiers; and

rendering in a first group medical codes corresponding to the medical finding; and

rendering in a second group separate from the first group medical codes corresponding to the medical recommendation,

wherein processing automatically the medical text in the radiology report further comprises generating a lexicon-based hierarchical decision tree for radiology by assigning each valued medical phrase to a location in the lexicon-based hierarchical decision tree.

20. The computer method of claim 19 wherein rendering in the first group medical codes corresponding to the medical finding comprises displaying on a screen the first group,

wherein rendering in the second group medical codes corresponding to the medical recommendation comprises displaying on the screen the second group.

21. The computer method of claim 19, further comprising transcribing dictated words into the medical text using a voice recognition and transcription system.

22. The computer method of claim 19, wherein the circuitry configured to determine which of the medical phrases are valued medical phrases comprises circuitry configured to determine which of the medical phrases are valued medical phrases including content having a corresponding value in a lexicon of radiological terms.

23. An apparatus, comprising:

hardware circuitry configured to process automatically medical text from a radiology report, the circuitry configured to process automatically medical text comprises circuitry configured to;

identify medical phrases contained within the medical text;

extract the medical phrases from the medical text;

determine which of the medical phrases are valued medical phrases, a valued medical phrase includes content having a corresponding value in a lexicon of relevant terminology in radiology, the valued medical phrase being either a medical finding or a medical recommendation;

reduce each valued medical phrase to one or more medical codes in radiology from at least one of Current Procedural Terminology, Fourth Revision (CPT4), International Classification of Diseases, Ninth Revision (ICD9), Systematized Nomenclature of Medicine (SNOWMED), a lexicon of radiological terms, or Unified Medical Language System (UMLS);

the one or more medical codes being medical identifiers; and

render in a first group medical codes corresponding to the medical finding; and

render in a second group separate from the first group medical codes corresponding to the medical recommendation,

wherein the circuitry comprises a processor,

wherein the circuitry configured to process automatically the medical text from the radiology report further comprises circuitry configured to generate a lexicon-based hierarchical decision tree for radiology by assigning each valued medical phrase to a location in the lexicon-based hierarchical decision tree.

24. The apparatus of claim 23, wherein the circuitry configured to render in the first group medical codes corresponding to the medical finding comprises circuitry configured to display on a screen the first group;

and

wherein the circuitry configured to render in the second group medical codes corresponding to the medical recommendation comprises circuitry configured to display on the screen the second group.

25. The apparatus of claim 23, further comprising circuitry configured to transcribe dictated words into the medical text.

26. The apparatus of claim 23 wherein the apparatus is configured to process automatically medical text in fourteen seconds from a thousand radiology reports in to one of reports comprising clinically significant findings or reports comprising recommendations for subsequent action.