Decision-support application and system for medical differential-diagnosis and treatment using a question-answering system

• US 10,515,073 B2
• Filed: 09/26/2017
• Issued: 12/24/2019
• Est. Priority Date: 09/24/2010
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bagchi, Sugato, Ferrucci, David A., Levas, Anthony T., Mueller, Erik T.
• Primary Examiner(s): Burgess, Joseph D
• Application Number:
  US15/715,527
• Publication Number:
  US 20180025127A1
• Time in Patent Office: 819 Days
• CPC Class Codes: A61B 34/10 Computer-aided planning, simulation or modelling of surgical operations
• CPC Class Codes: A61B 5/00 Measuring for diagnostic purposes radiation diagnosis A61B6/00; diagnosis by ultrasonic, sonic or infrasonic waves A61B8/00; Identification of persons
• CPC Class Codes: G06F 16/2428 Query predicate definition using graphical user interfaces, including menus and forms G06F16/2423 takes precedence
• CPC Class Codes: G06F 16/24522 Translation of natural language queries to structured queries
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/31 Indexing; Data structures therefor; Storage structures
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/532 Query formulation, e.g. graphical querying
• CPC Class Codes: G06F 16/90335 Query processing
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 3/048 Interaction techniques based on graphical user interfaces [GUI]
• CPC Class Codes: G06F 40/134 Hyperlinking
• CPC Class Codes: G06F 40/169 Annotation, e.g. comment data or footnotes
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06N 5/027 Frames
• CPC Class Codes: G16H 10/20 for electronic clinical trials or questionnaires
• CPC Class Codes: G16H 10/60 for patient-specific data, e.g. for electronic patient records
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof
• CPC Class Codes: G16H 40/20 for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/70 : for mining of medical data, e.g. analysing previous cases of other patients
• CPC Class Codes: G16H 70/00 : ICT specially adapted for the handling or processing of medical references
• CPC Class Codes: Y02A 90/10 : Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

A decision-support system for medical diagnosis and treatment comprises software modules embodied on a computer readable medium, and the software modules comprise an input/output module and a question-answering module. The method receives patient case information using the input/output module, and generates a medical diagnosis or treatment query based on the patient case information and also generates a plurality of medical diagnosis or treatment answers for the query using the question-answering module. The method also calculates numerical values for multiple medical evidence dimensions from medical evidence sources for each of the answers using the question-answering module and also calculates a corresponding confidence value for each of the answers based on the numerical value of each evidence dimension using the question-answering module. The method further outputs the medical diagnosis or treatment answers, the corresponding confidence values, and the numerical values of each medical evidence dimension for one or more selected medical diagnosis or treatment answers using the input/output module.

Claims

1. A method comprising:

receiving, by a decision-support system for medical diagnosis and treatment, patient case information regarding a patient, said decision-support system comprising a computerized device that has access to evidence sources containing medical domain knowledge;

analyzing, by said decision-support system, said patient case information to identify semantic concepts by recognizing phrases in said patient case information logically related to medical concepts;

generating, by said decision-support system, a query by automatically supplying said semantic concepts to blank slots in a query template to generate said query, said query comprising a medical question;

generating, by said decision-support system, answers to said query by applying natural language processing techniques to match said query to passages of said medical domain knowledge within said evidence sources, said evidence sources comprising information in unstructured form;

calculating, by said decision-support system, numerical values of evidence dimensions for each of said answers by assigning different numerical values to each of said answers based on the correctness of each of said answers for different semantic concepts;

calculating, by said decision-support system, corresponding confidence values for said answers by combining said numerical values of said evidence dimensions as calculated for said answer;

outputting, concurrently by said decision-support system to a decision-maker for said patient, said answers, said numerical values of said evidence dimensions for each of said answers, said corresponding confidence values for said answers, and at least one automatically generated link to said evidence sources for each of said answers so that said answers, said numerical values of said evidence dimensions for each of said answers, said corresponding confidence values for said answers, and said evidence sources for each of said answers are usable by said decision-maker to make a decision regarding said patient;

storing, by said decision-support system in a tangible repository, said answers, said numerical values of said evidence dimensions for each of said answers, and said corresponding confidence values for said answers;

in response to new medical domain knowledge being added to said evidence sources, automatically repeating, by said decision-support system, said generating of said answers to said query, said calculating of said numerical values of said evidence dimensions for each of said answers and said calculating of said corresponding confidence values for said answers in order to determine whether previously generated answers and corresponding confidence values stored in said tangible repository would change based on said new medical domain knowledge; and

,

automatically sending, by said decision-support system, an alert to said decision-maker based on a change occurring in any of said previously generated answers and corresponding confidence values, said alert being usable by said decision-maker to re-evaluate said decision.

2. The method according to claim 1, further comprising automatically repeating, by said decision-support system, said generating of said query, said generating of said answers to said query, said calculating of said numerical values of said evidence dimensions for each of said answers, and said calculating of said corresponding confidence values for said answers based on new patient case information being received.

3. The method according to claim 2, further comprising automatically sending, by said decision-support system, an additional alert to said decision-maker based on any additional change occurring in any of said previously generated answers and corresponding confidence values based on said new patient case information being received.

4. The method according to claim 1, said generating of said answers being carried out by analyzing medical domain knowledge content.

5. The method according to claim 1, said generating of said query being carried out by receiving said query in a form of at least one of a free-form query, a free-form statement, and keyword search.

6. The method according to claim 5, further comprising receiving, by said decision-support system, an initial query from said decision-maker, said generating of said query comprising:

in response to said initial query, analyzing said patient case information, to identify said semantic concepts; and

expanding said initial query using said semantic concepts to generate a specific natural language query.

7. The method according to claim 1, further comprising outputting, for a selected evidence dimension, each piece of evidence supporting said selected evidence dimension and an associated provenance for said piece of evidence.

8. The method according to claim 7, further comprising:

removing, by said decision-support system, a selected piece of evidence from said evidence sources contributing to a numerical value of an evidence dimension of a specific answer;

recalculating, by said decision-support system, said numerical value of said evidence dimension of said specific answer having said selected piece of evidence removed to produce a recalculated numerical value; and

recalculating, by said decision-support system, a new confidence value of said specific answer based on said recalculated numerical value.

9. The method according to claim 1, further comprising:

identifying, by said decision-support system, information relevant to said answers that is not contained within said patient case information as missing information; and

outputting, by said decision-support system, a request to said decision-maker to add said missing information to said patient case information.

10. The method according to claim 9, further comprising identifying, by said decision-support system, an amount said missing information affects corresponding confidence values of said answers.

11. The method according to claim 1, said alert being sent only when said change exceeds a difference threshold, said difference threshold being any of a percentage change in an answer, an answer polarity change, a percentage change in a confidence level and a confidence level polarity change.

12. A system comprising:

a first tangible repository maintaining patient case information regarding a patient;

a computer processor operatively connected to said first tangible repository and having access to evidence sources containing medical domain knowledge; and

a second tangible repository operatively connected to said computer processor, said computer processor receiving said patient case information from said first tangible repository,

said computer processor analyzing said patient case information to identify semantic concepts by recognizing phrases in said patient case information logically related to medical concepts, and generating a query by automatically supplying said semantic concepts to blank slots in a query template to generate said query, said query comprising a medical question,

said computer processor generating answers to said query by applying natural language processing techniques to match said query to passages of said medical domain knowledge within said evidence sources, said evidence sources comprising information in unstructured form,

said computer processor calculating numerical values of evidence dimensions for each of said answers by assigning different numerical values to each of said answers based on the correctness of each of said answers for different semantic concepts,

said computer processor calculating corresponding confidence values for said answers by combining said numerical values of said evidence dimensions as calculated for said answer,

said computer processor outputting, concurrently by said decision-support system to a decision-maker for said patient, said answers, said numerical values of said evidence dimensions for each of said answers, said corresponding confidence values for said answers, and at least one automatically generated link to said evidence sources for each of said answers so that said answers, said numerical values of said evidence dimensions for each of said answers, said corresponding confidence values for said answers, and said evidence sources for each of said answers are usable by said decision-maker to make a decision regarding said patient,

said computer processor further storing, in said second tangible repository, said answers, said numerical values of said evidence dimensions for each of said answers, and said corresponding confidence values for said answers,

said computer processor further, in response to new medical domain knowledge being added to said evidence sources, automatically repeating said generating of said answers to said query, said calculating of said numerical values of said evidence dimensions for each of said answers and said calculating of said corresponding confidence values for said answers in order to determine whether previously generated answers and corresponding confidence values stored in said second tangible repository would change based on said new medical domain knowledge, and said computer processor automatically sending an alert to said decision-maker based on a change occurring in any of said previously generated answers and corresponding confidence values, said alert being usable by said decision-maker to re-evaluate said decision.

13. The system according to claim 12, said computer processor further automatically repeating said generating of said query, said generating of said answers to said query, said calculating of said numerical values of said evidence dimensions for each of said answers, and said calculating of said corresponding confidence values for said answers based on new patient case information being received.

14. The system according to claim 12, further comprising a third tangible repository for maintaining medical domain knowledge content, said computer processor generating said answers by analyzing said medical domain knowledge content.

15. The system according to claim 12, said computer processor generating said query being carried out by said computer processor receiving said query as an input in a form of at least one of a free-form query, a free-form statement, and keyword search.

16. The system according to claim 12, said computer processor outputting a piece of evidence supporting a selected evidence dimension and an associated provenance for said piece of evidence.

17. The system according to claim 16, said computer processor removing a selected piece of evidence from said evidence sources contributing to a numerical value of an evidence dimension of a specific answer, recalculating said numerical value of said evidence dimension of said specific answer having said selected piece of evidence removed to produce a recalculated numerical value, and recalculating a new confidence value of said specific answer based on said recalculated numerical value.

18. The system according to claim 12, said computer processor identifying information relevant to said answers that is not contained within said patient case information as missing information and outputting a request to said decision-maker to add said missing information to said patient case information.

19. The system according to claim 18, said computer processor identifying an amount said missing information affects corresponding confidence values of said answers.

20. A computer program product comprising a tangible computer readable storage device storing computer readable program code comprising instructions executable by a computerized device, said computerized device having access to evidence sources containing medical domain knowledge and said computer program code comprising:

an input/output module receiving patient case information regarding a patient;

a patient case analysis module analyzing said patient case information to identify semantic concepts by recognizing phrases in said patient case information logically related to medical concepts;

a question generation module generating a query by automatically supplying said semantic concepts to blank slots in a query template to generate said query, said query comprising a medical question; and

a question-answering module generating answers to said query by applying natural language processing techniques to match said query to passages of said medical domain knowledge within said evidence sources, said evidence sources comprising information in unstructured form, said question-answering module calculating numerical values of evidence dimensions for each of said answers by assigning different numerical values to each of said answers based on the correctness of each of said answers for different semantic concepts,

said question-answering module calculating corresponding confidence values for said answers by combining said numerical values of said evidence dimensions as calculated for said answer,

said input/output module outputting, concurrently by said decision-support system to a decision-maker for said patient, said answers, said numerical values of said evidence dimensions for each of said answers, said corresponding confidence values for said answers, and at least one automatically generated link to said evidence sources for each of said answers so that said answers, said numerical values of said evidence dimensions for each of said answers, said corresponding confidence values for said answers, and said evidence sources for each of said answers are usable by said decision-maker to make a decision regarding said patient,

said input/output module further storing, said answers, said numerical values of said evidence dimensions for each of said answers and said corresponding confidence values for said answers in a tangible repository,

said question-answering module further, in response to new medical domain knowledge being added to said evidence sources, automatically repeating said generating of said answers, said calculating of said numerical values of said evidence dimensions for each of said answers and said calculating of said corresponding confidence values for said answers to determine whether previously generated answers and corresponding confidence values stored in said tangible repository would change based on said new medical domain knowledge, and

said input/output module automatically sending an alert to said decision-maker based on a change occurring in any of said previously generated answers and corresponding confidence, said alert being usable by said decision-maker to re-evaluate said decision.