Utilizing artificial intelligence for data extraction

• US 10,223,640 B2
• Filed: 02/09/2018
• Issued: 03/05/2019
• Est. Priority Date: 04/28/2017
• Status: Active Grant

• Current Assignee: Kyndryl Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: Martinez Hernandez Magro, Agueda, Carvajal, Herbert Barrientos
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Zidanic, Michael
• Application Number:
  US15/893,130
• Publication Number:
  US 20180314960A1
• Time in Patent Office: 389 Days
• Field of Search: None
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06N 5/045 Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
• 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 70/60 relating to pathologies

Abstract

Solved diagnosis case data is stored by utilizing a redundant discrimination net as a dynamic memory. The stored diagnosis case data is incorporated to form scientific descriptions within a medical knowledge base and heuristics within an empirical knowledge base. Diagnosis hypotheses are generated using an initial symptom description, the dynamic memory, and the medical knowledge base. The initial symptom description is received from an end user. A subset of the diagnosis hypotheses is created to form one or more solution cases. The one or more solution cases are presented to a subject matter expert. A diagnosis success or a diagnosis failure identifying, based on a response received from the subject matter expert, to form an assessed solution case. An assessed solution case is converted into experiences. The experiences are inputted into the dynamic memory. Data containing the assessed solution case is transmitted to a medical artificial intelligence analytics application.

Claims

1. A computer-implemented process for generating ailment, disorder, and disease diagnostic data for consumption by a medical artificial intelligence analytics application, the computer-implemented process comprising:

storing solved diagnosis case data utilizing a redundant discrimination net as a dynamic memory, wherein the solved diagnosis case data is in a form of an electronic document inputted by a subject matter expert, wherein the electronic document was ingested from the Internet by a natural language processor accessed by utilizing a web crawling application designed to locate medical data on the Internet;

incorporating, by utilizing automated reasoning, the stored diagnosis case data to form scientific descriptions within a medical knowledge base and heuristics within an empirical knowledge base, wherein the descriptions and heuristics include ailments, disorders, and diseases;

generating diagnosis hypotheses using an initial symptom description, the dynamic memory, and the medical knowledge base, the initial symptom description being received from an end user, wherein the medical knowledge base includes one or more cases extracted from medical journals and medical dictionaries, wherein the generating diagnosis hypotheses comprises;

receiving an initial symptom description from the end user, wherein the initial symptom description may be in the form of at least one of text, audio, or a video format;

transforming, utilizing natural language processing, at least one of text, audio, or a video format into data structures; and

comparing the data structures to substantially similar data structures stored in case memory;

transmitting, to the subject matter expert, the medical journals and medical dictionaries in the form of electronic documents;

creating a subset of the diagnosis hypotheses to form one or more solution cases;

presenting, after generating the diagnosis hypotheses based on the initial symptom description being received from the end user and after creating a subset of the diagnosis hypotheses to form the one or more solution cases, the one or more solution cases to a subject matter expert;

identifying, based on a response to the transmitted electronic documents and the presented solution cases received from the subject matter expert, a diagnosis success or a diagnosis failure to form an assessed solution case;

converting assessed solution case into experiences;

inputting the experiences into the dynamic memory, wherein the dynamic memory is a self-organizing process that incorporates new knowledge as a result and improves performance in future diagnostic case sessions;

transmitting data containing the assessed solution case to a medical artificial intelligence analytics application, wherein the transmitted data contains updated and validated solution case data;

incorporating, using a self-organizing process, new knowledge against existing scientific descriptions and heuristics;

proposing corrections and enhancements to the medical knowledge base by analyzing the new knowledge incorporated against existing scientific descriptions and heuristics, wherein the medical knowledge base maintaining module keeps the empirical knowledge base up to date with recent scientific and empirical data inputted by the subject matter expert, wherein the recent scientific and empirical data includes medical websites, scientific papers, published studies;

generating the updated and validated solution case data;

receiving a request, by an analytics system administrator, for the generated updated and validated solution case data according to a predetermined format criteria; and

exporting data containing the generated updated and validated solution case data based on to the predetermined format criteria to the analytics system administrator.