Please enable JavaScript.

Coggle requires JavaScript to display documents.

Data Science in Healthcare: Supporting Clinicians in the Assessment of…

- - - - clinical data #
        
        structured #
        
        EHR
        
        unstructured
        
        narrative clinical notes
    - - Deep Learning (DL)
        
        helpful to find hidden variables
  - - - questionnaires and clinical interviews
        
        can be biased
        
        can be subjective
        
        don't consider important variables
      - patient self-reports
        
        based on information that the patient is willing to share (not full picture)
      - evaluation during medical appointments
        
        patient’s mental state then and there
    - - NLP, ML and DL improved diagnosing
        
        ML (and DL) on structured clinical data
        
        NLP on unstructured clinical data
        
        game changer!
      - BUT!
        
        there are false positives
        
        results obtained on retrospective data
        
        they can differ in actual clinical practice
  - - - protect individuals’ rights to privacy and confidentiality
      - compliance with legal and ethical regulations
        
        on national level
        
        on international level
    - - The majority of research was undertaken in North America
        
        underrepresentation of racial and ethnic minorities in the training datasets
        
        possible biased outputs and discrimination
        
        how to apply it in different cultures?
        
        problem with stigma
        
        what about other variables relevant for specific cultures?
    - - human factors #
        
        resistance
        
        uncertainty
        
        clinician's evaluation vs. AI
        
        AI as a tool, not replacement of clinicians
      - non-human factors #
        
        infrastructure
        
        access to relevant patient data
      - both human and non-human factors #
        
        developing, deploying and maintenance
        
        distribution of resources depends on political decisions
        
        involving specialists from different fields
- - - - research undertaken under controlled conditions
      - results are not free from false positives
      - different challenges must be solved