Chapter 3. Development of a diagnostic AI-based assessment system using lexicogrammatical features

Chapter 3
Development of a diagnostic AI-based assessment system using lexicogrammatical features

This content is being prepared for publication; it may be subject to changes.

Article outline

3.1Overview
3.2Rationale for a supplemental language-based diagnostic tool
3.3Methods
- 3.3.1Corpus as training database and selection of individuals
- 3.3.2The texts
- 3.3.3Diagnostic differentiation by machine learning
  - 3.3.3.1Overview and rationale for machine learning approaches
  - 3.3.3.2Input
  - 3.3.3.3Output and experimental procedure
  - 3.3.3.4ASD differentiation model
    - 3.3.3.4.1ASD differentiation using tags
    - 3.3.3.4.2Differentiation using text
    - 3.3.3.4.3Differentiation using tag-and-text combinations
3.4Results
3.5Diagnostic performance and interpretive implications
- i.Medical transparency
- ii.Path to improved diagnostic accuracy
- iii.Cognitive insights
3.6Validity of text selection as a merit of the diagnostic model
3.7Limitations and future perspectives
- 3.7.1Verification process and methodological enhancements
- 3.7.2Analysis of false positives and false negatives
- 3.7.3Investigation of influences of comorbid conditions
  on lexicogrammatical choices
3.8Chapter conclusion