Methodology

TCMIO knowledge is built on a structured pipeline that ensures accuracy, traceability, and safety.

Content Pipeline

1. Structured Data: Core facts about herbs, formulas, symptoms, conditions, and patterns are stored in a relational database.
2. Fact Cards: Structured data is converted into verified fact cards with source links.
3. AI Draft Generation: AI generates explanatory content from structured data (never from model memory alone).
4. Citation Validation: Every claim is checked against linked sources.
5. Safety Check: The safety rules engine reviews content for risk level compliance.
6. Duplicate Detection: Content is checked for similarity with existing pages.
7. Quality Score: Content receives an automated quality assessment.
8. Publish or Review: Content meeting thresholds is published; borderline content enters manual review.

Evidence Classification

We distinguish between: Traditional Use, Modern Human Evidence, Animal or Laboratory Evidence, Safety Evidence, and Uncertain Evidence. Evidence status is rated as Insufficient, Early-stage, Limited, Mixed, Moderate, or Strong.