Clinical concept identification, extraction, and prediction system and related methods

Patent metadata

Field	Value
Publication number	US-10957433-B2
Application number	US-201916702510-A
Country	US
Kind code	B2
Filing date	Dec 3, 2019
Priority date	Dec 3, 2018
Publication date	Mar 23, 2021
Grant date	Mar 23, 2021

Abstract

A method includes the steps of determining a first concept from a text of a medical record from an electronic health record system, the first concept relating to a patient, identifying a match to the first concept in a first list of concepts, wherein the first list of concepts is not a predetermined authority, referencing the first concept with an entity in a database of related concepts, identifying a match to a second concept in a second list of concepts, the second list of concepts not directly linked to the first list of concepts except by a relationship to the entity, wherein the second list of concepts is the predetermined authority, and providing the second concept as an identifier of the patient's medical record.

First claim

To apprise the public of the scope of this invention, the following claims are made:

1. A method for generating structured records of a patient's health record, comprising:
receiving, at a processor, the patient's health record from an electronic health record system;
selecting, by the processor, an artificial intelligence model trained on electronic health records from one or more artificial intelligence models based on a type of document being evaluated by the processor;
determining, by the processor via an abstraction engine applying the selected artificial intelligence model, a first concept from a text of the health record, the first concept relating to the patient;
identifying a match to the first concept in a first list of concepts, wherein the first list of concepts is not designated as a predetermined authority and includes two or more tiers of concepts;
referencing the first concept with an entity in a database of related concepts;
identifying a match from the entity to a second concept in a second list of concepts designated as the predetermined authority and including two or more tiers of concepts, wherein the second concept is included in one of the two or more tiers of the predetermined authority, and wherein the predetermined authority is not directly linked to the first list of concepts except by a relationship to the entity;
retrieving, by the processor, a degree of specificity comprising selection criteria identifying one or more tiers of the two or more tiers of the predetermined authority, other than the tier including the second concept;
normalizing the second concept to a third concept included in the one or more other tiers, based at least in part on the second concept not satisfying the selection criteria;
providing the third concept as an identifier of the patient's health record, wherein the first, second, and third concepts and the identifier each have a respective category; and
generating, by the processor, a structured record from the identifier.

2. The method of claim 1 , wherein the respective category categories of the first concept, the second concept, the third concept, and the identifier are categorized as one of diagnosis, primary diagnosis site, metastatic diagnosis site, tumor characterization, standard grade, alternative grade, medications, surgical procedure, smoking status, comorbidities, adverse events, outcomes, performance scores, radiotherapy modality, radiotherapy units, imaging type, gene mention, immunology markers, TNM status, or AJCC stage.

3. The method of claim 1 , wherein the first list of concepts is one of a vocabulary, terminology, ontology, or code set.

4. The method of claim 3 , wherein the predetermined authority is another one of a vocabulary, terminology, ontology, or code set.

5. The method of claim 4 , wherein the first list of concepts is an ontology, the predetermined authority is an ontology, and the ontology of the first list of concepts does not directly link to the ontology of the predetermined authority.

6. The method of claim 5 , wherein the ontology of the first list of concepts not directly linking to the ontology of the predetermined authority signifies a lack of one of a hierarchical relationship or a non-hierarchical relationship and wherein a non-hierarchical relationship is a link comprising one of physically related to, spatially related to, temporally related to, functionally related to, or conceptually related to.

7. The method of claim 1 , wherein the entity in the database of related concepts is a concept in one of a vocabulary, terminology, ontology, or code set.

8. The method of claim 7 , wherein the entity has a first non-hierarchical relationship link between the entity and the first concept and a second non-hierarchical relationship link between the entity and the second concept.

9. The method of claim 8 , wherein the first non-hierarchical relationship link is a link comprising one of physically related to, spatially related to, temporally related to, functionally related to, or conceptually related to and the second non-hierarchical relationship link is a link comprising one of physically related to, spatially related to, temporally related to, functionally related to, or conceptually related to.

10. The method of claim 1 , wherein an identifier is a field having a structured format, the field encoding the second concept in the structured format.

11. The method of claim 1 , wherein the predetermined authority is one of a reference vocabulary, reference terminology, reference ontology, or reference code set.

12. The method of claim 11 , wherein the predetermined authority is categorized as one of diagnosis, primary diagnosis site, metastatic diagnosis site, tumor characterization, standard grade, alternative grade, medications, surgical procedure, smoking status, comorbidities, adverse events, outcomes, performance scores, radiotherapy modality, radiotherapy units, imaging type, gene mention, immunology markers, TNM status, or AJCC stage.

13. The method of claim 1 , wherein the first list of concepts is a medicine ontology, the entity is a medicine, and the predetermined authority is a reference medicine ontology.

14. The method of claim 1 , wherein the first list of concepts is a procedure ontology, the entity is a procedure, and the predetermined authority is a reference procedure ontology.

15. The method of claim 1 , wherein the first list of concepts is an outcomes ontology, the entity is an outcome, and the predetermined authority is a reference outcome ontology.

16. The method of claim 1 , wherein the first list of concepts is a medicine ontology, the entity is a treatment, and the predetermined authority is a reference diagnosis ontology.

17. The method of claim 1 , wherein the first list of concepts is a unified medical ontology, the entity is a medicine, and the predetermined authority is a reference medicine ontology.

18. The method of claim 1 , wherein the first list of concepts is a brand-name medicine ontology, the entity is a medicine, and the predetermined authority is a generic medicine ontology.

19. The method of claim 1 , wherein generating the structured record from the identifier comprises:
identifying a structured format for the identifier of the patient's health record;
extracting fields from the structured format;
assigning one or more values associated with the second concept to the extracted fields;
identifying one or more values from the text of the health record surrounding the first concept;
assigning one or more of the identified values to the extracted fields; and
providing the structured format having one or more assigned values as a structured record of the patient's health record.

20. The method of claim 1 , wherein the third concept is categorized as one of diagnosis, primary diagnosis site, metastatic diagnosis site, tumor characterization, standard grade, alternative grade, medications, surgical procedure, smoking status, comorbidities, adverse events, outcomes, performance scores, radiotherapy modality, radiotherapy units, imaging type, gene mention, immunology markers, TNM status, or AJCC stage.

21. The method of claim 1 , wherein the predetermined degree of specificity is based at least in part on metadata of the second concept.

22. The method of claim 19 , wherein the one or more values are extracted from sentences of the text using natural language processing (NLP).

23. The method of claim 1 , wherein the referencing the

Assignees

• Tempus Labs
• Tempus Labs Inc

Inventors

• Lucas Michael
• Ozeran Jonathan
• Taylor Jason
• Fernandes Louis

Classifications

• G16H10/20

  for electronic clinical trials or questionnaires · CPC title

• G06F21/6254

  by anonymising data, e.g. decorrelating personal data from the owner's identification · CPC title

• G16H10/60 Primary

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

• G06F40/30

  Semantic analysis · CPC title

• G06F40/274

  Converting codes to words; Guess-ahead of partial word inputs · CPC title