Sources and methodology
Editorial policy in full
1The three-axis system
Most services collapse everything into one axis: does it work or not. There are really three independent questions, and we never combine them into a single score.
Axis A · Evidence grade
| A | Established | Consistent efficacy in systematic reviews or multiple well-designed RCTs |
| B | Promising | A few RCTs exist but are small or inconsistent; or a review concludes "possible" |
| C | Preclinical only | In vitro or animal studies only. No human clinical evidence |
| D | Insufficient | Little or no human research. This means "not yet studied", not "does not work" |
| E | Evidence against | Well-designed studies show no effect, or harms outweigh benefits |
Axis B · Safety signal
| ▲ | Harm confirmed | Regulatory action or clear reports of human harm |
| △ | Caution | Contraindicated or requires caution in pregnancy, lactation, children, or specific conditions |
| † | Adverse-event literature exists | Literature on adverse events, toxicity or poisoning exists. Not a finding of harm — a pointer to what must be read |
| ‡ | Interaction literature exists | Interaction literature exists. This is a literature pointer, not an interaction checker |
| — | Not yet assessed | Modern safety sources (NCCIH, LiverTox) have not been consulted yet. This means not yet checked — not that nothing was found. |
| ○ | No signal found | No reports found within the sources searched. This is not a guarantee of safety |
Axis C · Mapping confidence
| ◇◇◇ | Analogical only | Conceptual analogy only. These must not be treated as equivalent |
| ◆◇◇ | Partial overlap | Only some symptoms overlap; the traditional concept is broader or narrower |
| ◆◆◇ | Strong correspondence | No official mapping, but symptom clusters and pathophysiology largely align in the literature |
| ◆◆◆ | Official code mapping | Official correspondence registered in KCD U-codes or the KTKP mapping dataset |
2How evidence grades are derived
- We run a per-item query against Europe PMC and count the pubType distribution (systematic review / RCT / preclinical).
- We check the number of registered studies on ClinicalTrials.gov.
- From those counts alone we propose a draft grade, shown as "unreviewed".
- A curator reads the abstracts before a grade is finalised. Grade E is never assigned automatically.
- The exact query is published with each item so you can reproduce it.
3Data sources · Licence notices
This table is generated from the source registry in code, so the documentation cannot drift from the implementation.
| Source | Licence | Redistribution | Commercial |
|---|---|---|---|
| Europe PMC REST | 논문별 상이 (응답의 license 필드로 판별) | Metadata only | conditional |
| ClinicalTrials.gov API v2 | Public Domain (U.S. Government work) | Full redistribution | yes |
| OpenAlex | CC0 | Full redistribution | yes |
| NCCIH Herbs at a Glance | Public Domain (복제 권장) | Full redistribution | yes |
| KTKP Korean–Western disorder mapping (15059048) | 공공누리 제4유형 (출처표시 · 상업적 이용금지 · 변경금지) | Metadata only | no |
| KTKP materia medica (15002148) | 이용허락범위 제한 없음 (공공누리 유형 미표시) | Full redistribution | conditional |
| KTKP formulas (15057537) | 이용허락범위 제한 없음 (공공누리 유형 미표시) | Full redistribution | conditional |
| LiverTox | Public Domain | Full redistribution | yes |
| NIKOM/NCKM Clinical Practice Guidelines | 저작권법 보호 저작물 — 조건부 무료 열람·이용 (일괄 재배포 허락 아님) | Link only (no mirroring) | no |
| World Flora Online Plant List | CC0 | Full redistribution | yes |
| PubChem PUG-REST | NCBI 무저작권 주장 | Full redistribution | yes |
| WHO ICD-11 | CC BY-ND 3.0 IGO | Metadata only | conditional |
| KCD Korean Medicine U-codes | 통계청 고시 (공공 분류체계) | Full redistribution | yes |
| Korean Medicine Classics DB | KIOM 저작권 — 재가공·유사서비스·AI 학습 금지 | Link only (no mirroring) | no |
| MSK About Herbs | All rights reserved — 데이터 추출·마이닝 일체 금지 | Link only (no mirroring) | no |
| Korea Traditional Knowledge Portal | robots.txt 에서 일반 크롤러 전면 차단 | Link only (no mirroring) | unknown |
| KIOM OASIS | KIOM 저작권, API 없음 | Link only (no mirroring) | no |
| Cochrane Library | 구독 기반, API 없음 | Link only (no mirroring) | no |
| local resources/ — LLM-generated English edition of Donguibogam scans (unverified) | 프로젝트 소유 (원저작물은 퍼블릭 도메인) | Full redistribution | yes |
4Editorial policy
- Absence of evidence is not evidence of absence. Grade D is never written up as "does not work".
- Historical and modern claims are grammatically separated: "the text describes…" versus "a 2023 systematic review reported…".
- Every evidence claim carries a clickable source. Unsourced efficacy statements are not published.
- Safety information is always shown, independent of the evidence grade.
- A disclaimer appears on every detail page.
- Automatically derived grades are always labelled "unreviewed".
- We neither advocate for nor dismiss traditional medicine. We show the state of the evidence.