Our Methods

How we create clinically grounded content.

Content Development

1

Taxonomy curation

We maintain a taxonomy of clinical topics based on DSM-5-TR categories, clinical screeners, and common symptom queries.

2

Data grounding

Each article is informed by clinical data including DSM criteria, evidence confidence scores, and screening alignment data where available.

3

LLM-assisted writing

Articles are generated by large language models following strict clinical writing guidelines, then validated for structural completeness, citation presence, and non-diagnostic language.

4

Automated validation

Content passes automated checks for structural completeness, factual consistency, citation presence, and adherence to writing standards.

Writing Standards

  • College reading level - accessible but clinically precise
  • Non-diagnostic language - "patterns consistent with" rather than "you have"
  • DSM-5-TR criteria cited where applicable
  • Prevalence data from DSM-5-TR or NIMH estimates (no fabricated statistics)
  • Clear distinction between established findings and emerging research
  • Every article includes "when to seek help" guidance

Quality Controls

  • Articles contain 5-10 substantive sections covering the topic in depth
  • FAQ sections target real search queries in a People Also Ask format
  • Each article includes 3-6 citations from peer-reviewed or clinical sources
  • Content is grounded in existing clinical databases when available
  • Medical disclaimers are included on every clinical page

Limitations

Our content has inherent limitations:

  • Content is generated by Claude Opus 4.6 and has not been reviewed by a licensed clinician
  • Content may not reflect the most current research at any given time
  • Individual clinical presentations vary - our content describes general patterns, not individual cases
  • This resource is not a substitute for professional clinical assessment