Research

ES, version 2022-04-11 /

• Introduction:
  • The quality of the knowledge graph database depend on research.
• Objectives:
  • Improvement of the Knowledge graph database.
  • Attempts to discover new patterns
• Approaches:
  • Synthesis of knowledge:
    • Creation and maintenance of the knowledge base.
    • The users of the decision support modules expect the best possible synthesis of the knowledge. Usually they would not delve in scientific discussions.
    • The research team must follow the evolution of scientific knowledge in new publications. When necessary some aspects of the knowledge base must be adapted. This maintenance should be reviewed by trusted scientific associations in diverse specialized domains.
  • Sources of knowledge bases:
  • Review of available knowledge bases, particularly when available in the public domain.
  • Particular attention for types and weights of relationships:
    
    • Given a symptom, likelihood of several Problems.
    • Given a disease, relative frequencies of symptoms, with sensivity and specificity.
    • Effectiveness of treatments
    • ...
  • Analysis of follow-up of patients:
  • Own results in patient follow-up:
    • Not yet any results.
    • Intended to improve decision performance over time.
    • Research about the result of recommended actions:
      • How far have our recommendations been followed ?
      • Evaluation of the results of our recommendations ?
        • Criteria will be defined, for example quality of life after 3 months ? survival rates ? ..
  • Analysis of other patient data bases available today:
    • MIMIC:
      • A large data base from the "Beth Israel Deaconess Medical Center" (BIDMC), https://physionet.org/content/mimiciv/0.4/ .
      • Based on real data of 40.000 admitted patients in intensive care.
      • Access for research purposes is allowed on condition to comply with credential requirements.
  • Seek unsuspected new patterns:
    • Exploration of the new possibilities of graph software tools:
      • ....
  • ............
  • ......
• Evolution:
  • Current results:
    • Exercises have already been done, selective retrievals in the knowledge base.
    • Analysis of concepts and relationships in a few knowledge bases already converted in graph.
  • Next steps:
    • More focus on analysis of the types of relationships.
    • The weight of relationships will be improved in function of observed results.
      
• Links to more documents:
  • ...........
  • ....
• Contact persons:
  • ....